RU2511598C2 - Method of detecting random low-energy signals - Google Patents

Abstract

FIELD: physics, radio.

SUBSTANCE: invention relates to radio engineering and can be used for purposes of radio control, radio monitoring and detection of random low-energy signals. The method is based on analysing fractal properties of received signals. According to the invention, a random low-energy signal is detected by estimating the Hurst exponent and comparing it with a threshold. The method enables to obtain accurate results with signal-to-noise ratio of the processed signal of about minus 10 dB.

EFFECT: low threshold of detecting random low-energy signals.

2 dwg

Description

The invention relates to radio engineering and can be used for the purposes of radio monitoring, radio monitoring, and the detection of random low-energy signals. The method is based on the study of the fractal properties of the received signals.

Currently, various methods are known for detecting random low-energy signals (see patents of the Russian Federation No. 2429494, 2359406, 2285274). Known detection methods are based on the use of two-channel detection using consistent filtering, calculating the correlation coefficient, amplitude spectrum, calculating the detection threshold, comparing the amplitudes of the spectral components with the detection threshold.

There is a method for detecting pulsed radio signals in the presence of signals of interfering reflections and white noise. The claimed invention relates to the field of radio engineering and consists in the use of two-channel detection, in which in the first main detection channel a coordinated signal filtering is applied with a threshold decision on its presence or absence according to the selected criterion, and in the second additional detection channel, the time from the reference the subsequent measurement of the noise voltage in it and the calculation of its correlation coefficient between the reference and information samples, on the basis of which, when tnoj power output noise P _w of the matched filter and the target false alarm probability P _p value is calculated decision threshold, which is set to control the threshold device further detection channel, the final decision on the presence or no signal take on the basis of particular solutions of fundamental and supplemental channels detection by the rule: a signal is detected if at least one of the private detection channels is detected, incoherent detection and mayutsya of envelope process on output of the linear envelope detector, so at the time of the reference frame measured value of the noise envelope U _u (τ ₀₎ and the calculated envelope value of the correlation coefficient R ₀ noise at the matched filter output between the reference and information samples for a given value threshold P (τ ₀ ) [Patent RU 2285274 C2, METHOD FOR INCORRECT DETECTION OF A PULSE RADIO SIGNAL ON THE BACKGROUND OF AN INTERFERRING RADIO PULSE AND WHITE NOISE, published on 10.10.2006].

The disadvantage of this method is that it does not allow the detection of random low-energy signals and is effective only in the case of a priori certainty regarding the type of received signal and at a high signal to noise ratio of 40 dB.

Also known is a method of detecting ultra-small radio signals based on the use of two channels with the same resonant circuits and a subtracted process. The resonant circuits at time to lead to an excited state by briefly and simultaneously supplying voltage, a greater rms voltage of noise and interference in them. At the same time, both channels provide a radio signal from the antenna. The phases of the input signals of the resonant circuits are successively changed by introducing phase shifts φ ' ₁ and φ' ₂ with opposite signs into the channels. The upper envelopes of the total signals of the resonant circuits — S _A1 (t) and S _A2 (t) _—are distinguished. The difference of these envelopes is determined - ΔS _A (t). When the envelope difference reaches its maximum value - ΔS _{A, the} introduced phase shifts φ ' ₁ and φ' ₂ are fixed. After that, a useful signal is formed with the amplitude x = ΔS _A / 2 (1-exp (-λt)), where λ is the attenuation coefficient of the loops, and the phase, defined as the arithmetic mean of the phase shifts φ ' ₁ and φ' ₂ , taken with the inverse the sign [Patent RU 2359406 C2, METHOD FOR DETECTING ULTRA SMALL RADIO SIGNALS AND A DEVICE FOR ITS IMPLEMENTATION, published on 06/20/2009].

The disadvantage of this method is that it is effective only in the case of a priori certainty regarding the type and parameters of the received signal. In addition, this device is technically difficult to implement, since constant excitation of the resonant circuits and the supply of a useful signal from the antenna at certain times are required. There is also no information on the threshold for detecting low-energy signals.

Closest to the proposed method is a method for detecting a plurality of narrow-band radio signals in a wide frequency band, which consists in performing analog-to-digital conversion of a wide-band input process (ballscrew), accumulating complex ballscrew samples and sequentially generating samples with ballscrew samples, calculating the amplitude spectrum of the generated sample with samples Ballscrew, calculating the detection threshold for each component of the amplitude spectrum as the sum of the smoothed spectrum estimate in the moving the frequency axis of the weighing window and the function of their standard deviation, comparing the amplitudes of the spectral components with the detection threshold, deciding on the detection of radio signals in the studied frequency band based on the majority processing of the stored results of comparing the amplitudes of the spectral components with the detection threshold obtained by analyzing six sequentially taken samples with ballscrew readings, output of detection results. The technical result consists in reducing the influence of the signal-noise situation [Patent RU 2429494 C1, METHOD FOR DETECTING MULTIPLE VISIBLE RADIO SIGNALS IN A WIDE BAND, published on 09/20/2011]. This invention is selected as a prototype.

The disadvantage of this method is the need to perform a large number of calculations, including estimating the amplitude spectrum of the input process, calculating the detection threshold for each component of the amplitude spectrum and comparing the amplitudes of the spectral components with the calculated detection threshold, which requires significant computational resources and processing time. In addition, to carry out the estimates, it is necessary to accumulate several samples to make a majority decision, which requires additional processing time; this is unacceptable when processing random signals that can appear briefly and at random times.

The present invention is to detect random low-energy signals and eliminate the disadvantages of the prototype is to simplify signal processing and reduce computing costs. This problem in the proposed method is solved by applying the processing of harmonic signals using their fractal properties, as shown in the block diagram in figure 1.

The choice of fractal processing is due to the fact that fractal processing allows us to describe the integral characteristics of signals that are independent of the moment of arrival and are calculated for the entire duration of the signal. The digitized signal is a time series of data. The Hurst exponent was chosen as an indicator characterizing the fractal properties of time series. To determine the Hurst exponent, the normalized swing method is used.

The principle of operation of this method is as follows. The sums of deviations are calculated for different time periods and for a different number of consecutive observation moments that act as the measurement scale. Let X = {X _n } be a given set of values of the analyzed time series (signal). Then the accumulated deviation from the average for a certain interval is defined as:

$$\Delta (t)={\displaystyle \sum _{i=\mathrm{one}}^t\left(X_i-\overline{X}\right)},(\mathrm{one})$$

- среднее выборочное по ряду в целом.Where $$\overline{X}$$

- the average sample for the series as a whole.

 Accordingly, the span R (t):

$$R(t)=\max [\Delta (t)]-\min [\Delta (t)],(2)$$

Where

Δ (t) is the accumulated deviation from the average for a certain time interval t,

that is, the difference between the maximum and minimum deviations from the average, this characteristic differs from the difference between the maximum and minimum values of the time sequence.

To describe and compare different time series, for example, for signals with different types of modulation, the normalized dimensionless characteristic, the normalized range, is more convenient.

The normalized range of the interval is determined as follows:

$$\frac{R(t)}{S(t)}=\frac{\max [\Delta (t)]-\min [\Delta (t)]}{S(t)}=\frac{\max [\Delta (t)]-\min [\Delta (t)]}{\frac{\mathrm{one}}{n-\mathrm{one}}{\displaystyle \sum _{i=\mathrm{one}}^t{(X_i-\overline{X})}^2}},(3)$$

where S (t) is the standard deviation. According to Hurst’s formula:

$$\frac{R(t)}{S(t)}={(at)}^H,(\mathrm{four})$$

where a and H are some constants.

. от ln(t). Таким образом, параметр Н можно оценить, изобразив график указанной зависимости, и, используя полученные точки, подобрать по методу наименьших квадратов прямую линию с наклоном Н.The constant H can vary in the range from 0 to 1 and is called the Hurst exponent. It characterizes the slope of the linear graph $$\ln \left(\frac{R(t)}{S(t)}\right)$$

. from ln (t). Thus, the parameter H can be estimated by plotting the indicated dependence, and using the obtained points, select the straight line with the slope of N. using the least squares method.

There are three different classifications for the Hurst exponent:

1) H = 0.5 - indicates a random series, events are random and not correlated.

2) 0≤N <0.5 - this range corresponds to a harmonic signal.

3) 0.5 <Н≤1 - corresponds to an increasing or decreasing row.

When propagating, radio signals are exposed to white noise. Accordingly, with a decrease in the signal-to-noise ratio, the Hurst parameter increases, approaching the value of 0.5.

To test the proposed method for detecting random low-energy signals, we used the types of modulation widely used in space communication systems: FMN-2, FMN-4, FMN-8, KAM-16. The dependence of the value of the Hurst indicator on the signal-to-noise ratio for signals with various types of modulation is shown in Fig. 2. As can be seen from figure 2, when the signal-to-noise ratio is less than -15 dB, the Hurst exponent becomes 0.5 for all types of modulation, that is, the Hurst exponent characterizes the process as completely random. With an increase in the signal-to-noise ratio, the Hurst exponent decreases, and at large values of the signal-to-noise ratio, it stabilizes.

It has been found experimentally that the use of the Hurst parameter as a detection criterion makes it possible to efficiently detect random low-energy signals whose signal-to-noise ratio is of the order of minus 10 dB.

Achievable technical result of the proposed method is to reduce the detection threshold of random low-energy signals. This result was obtained by applying the processing of harmonic signals using their fractal properties. The proposed method provides a result and when exposed to noise on the signal under study (see figure 2). The threshold value of the signal-to-noise ratio at which the claimed result is executed is minus 10 dB. In this case, the signal processing is carried out in the mode as close as possible to real time.

Claims (1)

A method for detecting random low-energy signals based on analog-to-digital conversion of the input process, the results of which are sampled, characterized in that the Hurst parameter is used as a detection criterion, the value of the Hurst indicator is determined from the processing results, if the value of the Hurst indicator is in range from 0 to 0.5, then decide on the detection of the signal.

Images