RU2491717C2 - Method of increasing signal-to-noise level (ratio) using "disturbance damping principle" - Google Patents

Abstract

FIELD: information technology.

SUBSTANCE: method of increasing signal-to-noise level (ratio) in spatial selection of signals is characterised by that it includes an operation for synchronisation thereof and then summation of all components of the same signal source, received at spaced points, mutual spatial coordinates of which should be located in a region whose maximum dimensions are much larger than the maximum wavelength of the received wavelength range. Another method of increasing the signal-to-noise level (ratio) with variation of frequency-time parameters of the signal includes an operation for synchronising frequency-time elements of the input signal and then summing all frequency-time elements of the input signal formed at the transmitting side and for which frequency-time parameters of noise realisations, received with the signal in the same band, must satisfy independence conditions for random values.

EFFECT: extracting information during reception at noise levels exceeding the signal level.

7 cl, 14 dwg

Description

FIELD OF TECHNOLOGY

The invention relates to the field of reception, both electromagnetic signals (including optical), and for signals propagating as a result of elastic vibrations of the medium (acoustic and hydroacoustic) in which they propagate, regardless of the frequency range. Reception of signals is carried out in astronomy, radio engineering, acoustics, sonar and other areas of human activity, both for the purpose of extracting some information carried by the signal, and for the purpose of measuring the parameters of the signal itself. This determines the technical field to which the claimed solution belongs. It is assumed in connection with the stated decision that reception can be carried out not only against the background of noise of a natural or technogenic nature, but also deliberately introduced into the structure of the transmitted signal.

BACKGROUND

There are several close analogues of the proposed solution. All of them, to varying degrees, contain the beginnings of a new solution.

The first analogue: A system of diversity reception on two antennas, which increases reception efficiency by 3 ... 5 dB, which is equivalent to a 2-fold increase in the power of a subscriber station. [one]. The main purpose of the use of the diversity reception system is to combat signal fading resulting from multipath propagation, therefore, the positive effect that has appeared is to increase the reception efficiency by 3 ... 5 dB, used as the main result in the claimed invention, has not been further developed in this analogue.

The second analogue: phased array antenna (PAR, the most advanced analogue of the radar), where up to a thousand input signals are summarized. A useful signal is extracted by spatial selection. Both the emitted pulse and the reflected signal of monochrome frequency are synchronized in phase and not in time delay, as in the claimed invention, before summing for the final gain [2].

The task of isolating a useful signal from noise is not set here. Probably, this task here is carried out under certain conditions, but not because of the conscious and deliberate use of this result.

Third analogue: a communication system consisting of many EMR receivers, in which of all the signals, one of the strongest (and, therefore, the most reliable) of the received signals is automatically selected (and switched to the output). The task of isolating a useful signal from noise here is also not posed as feasible. The literature describes many ways to select according to various criteria the most optimal receive channel, including the addition of signals from several channels [3]. But the result of increasing the reception range and increasing the reliability of the received information here is achieved by a simple engineering solution. Closest to claim 1 of the claims is RF Patent No. 2075832 of 03/20/1997 [4].

Fourth analogue: A parabolic antenna, which is defined as the geometrical location of points equidistant from the focus and aperture plane of the antenna. This ensures that the field is in phase in the aperture of the antenna when it operates as an emitter and the addition (amplification) of the signal in focus, regardless of the frequency and phase when receiving signals. Its disadvantages include: 1) the need for mechanical rotation to change the direction of reception, and 2) the lack of flexible selection of algorithms for spatial selection of signals.

The fifth analogue (according to paragraph 3 of the claims). The literature describes various methods of suppressing interference to a radio reception. As theoretically ideal, but practically difficult to implement, the compensation method is given [5]. In practice, this method of noise suppression is effective only when additive interference with parameters regularly changing in time prevails in the stream. It is used, for example, to suppress certain types of industrial and atmospheric interference. In other cases, other, as relatively simple, methods are used: frequency filtering, temporal selection and amplitude limitation, based on the difference in signal and noise, respectively, in the frequency spectrum, arrival time and level; and methods that are difficult to implement, for example: synchronous detection, correlation reception, and also use receiving antennas with a narrow radiation pattern, which is most advisable when the signal and interference "arrive" to the antenna from different directions.

The closest in essence analogue to paragraph 3 of the invention (as a way to combat interference) is synchronous detection, in which a useful signal is fed to the input of the terminal adder after multiplication with a certain reference signal, and fragments of the interference after multiplication with the reference are at the input of the terminal adder in out of phase, due to which they are mutually compensated. Expressing it differently, due to the non-coincidence of the interference frequency and the standard, in contrast to the signal, its spectrum after multiplication shifts upward in frequency relative to the signal, which makes it easier to suppress it (interference) [6]. Due to the presence of a terminal adder as part of a synchronous detector, a phenomenon occurs in it, described by the principle of noise attenuation [7]. On the other hand, the need to isolate the above-mentioned reference signal from the input signal implies its sufficient level compared to the level of interference received at the receiver input with the signal. This makes it impossible to isolate a useful signal from an interference mixture when the signal level is insufficient or less than the interference level.

SUMMARY OF THE INVENTION

In the radio range formula [8], the ratio of signal power to noise power is key. At a given wavelength, antenna gains cannot provide decisive advantages. The ratio of the transmitter power to the power at the receiver input, hereinafter referred to as signal-to-noise (S / N), allows, from a mathematical point of view, to vary the radio range over a wide range.

Historically, all technical solutions in the field of signal reception, whether it is a radio station signal, a laser beam at the optical fiber output, a reflected radar signal or an echo sounder signal, a signal that has come down to us from a distant star, and so on, have been aimed at increasing the sensitivity of devices receiving these signals. But the creators have always come across a certain limit when, due to external or internal noise inherent in the devices, the signal became impossible to receive with the right quality. The signal-to-noise level was below the critical level [9]. Various ways have emerged to overcome this barrier. One of them is the claimed technical solution.

For reliable reproduction of the transmitted information, not only the input S / N ratio plays a role, but also the ability of the decoder (demodulator) to extract information at a given S / N level. Given the existence of a theoretical limit for the S / N ratio, at which it is still possible to extract something from the received signal for a given communication channel bandwidth, a solution to increasing the communication range should be sought between the receiver input and the decoder input in the field of mathematical methods, methods and processing algorithms received signal. One such solution is the claimed invention.

Traditionally, the design "signal source - communication channel - receiver" is a one-dimensional structure. According to the ubiquitous structure of the communication system, the signal is mixed with noise in the communication channel. But in fact, in the vast majority of cases, the signal and interference (C + W) fall into the antenna, which, in fact, is the input of the receiver from different directions. That is, the mixture of signal and noise at the input of the receiver has a spatio-temporal structure. The proposed patent solution just uses this property of the input mixture C + N.

The attenuation phenomenon to which both the noise and the signal is exposed in the claimed solution is described from a practical point of view by the “interference damping principle” (PZP) and is described below. According to the classification of anti-interference methods, cited by V.A. Kotelnikov in his classic work "Theory of potential noise immunity", according to claim 1, 3 of the claims, the first two methods are used simultaneously, this is increasing the signal power and reducing the noise level.

The term "spatial selection" used in claim 1 of the claims is understood here by analogy with the well-known term "temporary selection", as a physical phenomenon that occurs when many spatially separated reception points are used for reception. To obtain amplification and the occurrence of spatial selectivity, signals from the receiving points are fed to the input of the summing device in the same phase, which is determined by the same wavefront. So in a parabolic antenna, the synchronization of the delay time of the wave front is achieved by geometric means. All other signals that fall into focus after reflection as a result of the non-ideality of the reflecting surface are noise, and they are repaired by the principle of the PPP.

The figure 1 shows, as an example, obtained by numerical simulation the characteristic of selectivity. The x-axis indicates the offset angle between the direction to the radiation source and the direction to the receiving point, the step for calculation was 6 degrees. On average, signal attenuation is 7 times (≈17 dB). The antenna field includes 37 receivers located radially relative to the center at approximately the same distance from each other. Already at the first deviation of 6 degrees, the total signal fell to the average of the minimum level values.

In modern literature, two types of interference are distinguished, additive and multiplicative. The total signal is described by the expression:

Where, Sp - signal in the mixture with interference before processing at the receiving device;

Ar is an additive noise, its sources are internal and external noise;

Mr - a multiplicative interference that occurs as a result of nonlinear processes and is not observed without a signal, most often it is a modulation type.

For the modulating nature of Mr, expression (1) is rewritten in the form:

Where, m is the modulation depth coefficient.

Since in practice it is impossible to completely eliminate Mr, and its reduction to zero will lead to the disappearance of the signal according to expression (1), then to describe the signal / noise mixture it is more correct to use expression (2), which for a negligible level of Mr or m reduces to the known expression

As a result of synchronization of the summarized useful signals when receiving from the far zone, when the difference between the minimum and maximum value of the signal S _i becomes negligible, from (1) and (2) we obtain expressions (3) and (4):

In both expressions given above, the first terms on the right-hand side of equalities (3) and (4), in accordance with the principle of the PZP, increase significantly slower than the second sums with increasing n, due to which the components containing the useful signal are amplified. The expressions in parentheses containing the sums of Mr in both equalities, similarly to the sum of additive noise with increasing n, increase much more slowly than the product n * S. Moreover, in expression (3), with an increase in the level of Mr, the fraction of Ap in the signal decreases and the dependence of S on n is lost, which also raises the assumption that the signal-to-noise representation does not correspond to expressions (1) and (3).

In the following presentation, everywhere, by reducing the level of interference during summation, we should understand its reduction relative to its maximum possible level, in contrast to what happens with a useful signal, which grows in proportion to the product of n by S in expression (4) due to synchronization. Synchronization can be performed either only by the carrier frequency (according to claim 1 of the formula), or, as shown in one example implementation, by the envelope (by the modulating signal).

The term PPP is introduced to denote a phenomenon that is associated with the fact that when summing the nth number of independent random variables, the law of the probability distribution of the deviation of the random variable from its average value changes, which in the limit converges to the normal law. ¹ ( ¹ According to Proposition 1, when summing random variables, their probability distribution law changes.)

In addition, the independence condition can be satisfied both in the case of taking n samples from the received signal sequentially in time, and in the case of taking n spatially separated samples of the same signal. ² ( ² Claim 2, the confirmation of which is given in the next section of this description.) This leads to the presence of 4 points in the claims. The boundaries for fulfilling the independence conditions are given below.

The description below shows that a change in the law of the probability distribution of a deviation of a random variable from its average value when summing the nth number of independent random variables is also clearly manifested with a different probability distribution of the summed random variables, which also converges to normal law ³ with an increase in the number of terms. ( ³ Claim 3, an example of confirmation is given in the next section of the description of the invention.)

Further, also shows that the relative level of noise (amount Api and Mpi the right side of Expression (3), (4) divided by u and the maximum deviation of noise) in accordance with the principle of increasing the number of terms asymptotically approaches a certain minimum limit ⁴ . ( ⁴ Statement 4 states the asymptotic decrease in the relative noise level with an increase in the number of summed samples.)

The four given Statements, marked with underlining and footnotes, together constitute the basis of the principle of interference attenuation.

Theorem A.M. is known in mathematics. Lyapunov, formulated and proved by him in 1901, which claims that whenever a random variable is formed as a result of adding a large number of independent random variables with finite variances, the distribution law of this random variable turns out to be a practically normal law. [10]. This probability theory theorem can and / or should be considered the basis of Propositions 1 and 3 of the PPP.

According to the "Law of Large Numbers", which unites the group of theorems of probability theory, which is the basis of Proposition 4, under certain fairly general conditions, with an increase in the number of summed random variables, the arithmetic mean of their sum tends to the arithmetic mean of mathematical expectations and ceases to be random at infinity. The Laplace integral theorem, in turn, defines the law of variation of the variance of the sum obtained from the number of terms of random variables, which turns out to be proportional to the square root of the number of terms. [eleven].

Setting the emphasis in the title of the PPP on the main property of this phenomenon - the property of damping, the introduction of Statement 2 and the transfer of the three above-mentioned laws from the field of probability theory, which describes the properties of random variables, into the region describing the laws of wave propagation. through which the transfer of information is carried out create the PPP tool, which underlies the invention.

Values of the correlation function are considered to be a quantitative measure of the independence of random samples or random processes, on which the noise decay rate depends. So, samples separated by time intervals that are multiples of 1 / (2F) turn out to be mutually uncorrelated, which means independence for Gaussian values [12], where F is the upper frequency of the spectrum of the signal band. At the same time, the values of random processes are uncorrelated only with an unlimited frequency band. Any limitation of the frequency band introduces a certain correlation into the process, and only process values that are at least 1 / 2F apart from each other can be considered independent of each other [13]. So, according to expression (4), the correlation coefficient or the value of the correlation function for the useful signal tends to the maximum value. Without synchronization, the useful signal undergoes the same attenuation as interference.

To confirm with the help of computational models (VM) Claims 1 of the PPP, the summed random variables are taken subordinate to a random law with a uniform (rectangular) probability distribution law. Such a probability distribution is close to white noise, is used in a computer software application to generate random numbers, and differs from it in its finite spectrum. [fourteen]. Another assumption in constructing all VMs is to operate with discrete probability distributions and select the scale of the dynamic range of signals from a series of integers [15]. We also assume that the noise in any of the n reception channels has the same dynamic range.

Another assumption is associated with the transition in all VMs from the signal power or interference to the linear characteristic referred to in radio electronics as the average straightened value (SVZ in mathematics is called the arithmetic mean of the values taken modulo the values) and is due to several reasons:

1. According to [13], the power of a discrete signal with zero mathematical expectation is calculated by the formula:

где U_i(t) - i-я выборка сигнала.

where U _i (t) is the ith signal sample.

что приводит к сокращению большого объема операций. Операция возведения в квадрат не несет никакой информации и СВЗ является линейной характеристикой относительно значений выборок.At the same time, the SVZ is equal to

which leads to a reduction in the large volume of operations. The squaring operation does not carry any information and the SVZ is a linear characteristic with respect to sample values.

2. On the simplest models for a harmonic signal, a random signal, an amplitude-modulated (AM) signal and narrow-band interference (SCP), it was found that:

Moreover, the constant does not depend on the initial amplitude; when the carrier frequency changes, its value changes insignificantly only in the third decimal place, and depends linearly in the last two cases on the modulation depth m.

- harmonic signal - Const is 1.236;

- random signal - Const is 1.32; Fluctuations in re-calculations less than 3%;

- AM - signal (Signal) and soft starter (Rothepa) depend on m very close.

m 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 Signal 1.3101 1,3334 1.3598 1,3886 1.4208 1.4567 1,4954 1,5372 1,5821 Pomeha 1,3104 1,3334 1.3586 1.3879 1,4307 1.4571 1,4995 1,258 1,5904

In all VMs, both the signal and the interference were modulated with a modulation coefficient m of grass 0.5, while calculating the S / N ratio, the constants in the numerator and denominator are reduced, which proves the validity and expediency of this assumption.

Getting to the description of VM, confirming the validity of the claims of the PPP, we introduce a number of notation. To calculate the change in the probability P (x) of the deviation of a random variable (interference) x from its average value, we set the dynamic range of its maximum deviation D / 2. The number of all possible states N when summing n independent random variables (the number of receive channels) is determined by the expression:

It can be seen that the number of states is equal to twice the dynamic range raised to the power of the number of summable independent random components.

The number of all possible sums C (the number of events) is determined from the equality:

The smallest and largest sums are respectively equal for the case when the interference changes in the positive and negative region and its distribution is symmetrical with respect to zero,

For a positive region in the range from 1 to D, they are respectively equal:

The probability of a specific sum P (x) equal to P (x, n) and P (C i) will be determined by the number of all possible combinations giving the sum C i divided by the number of all states N. Accordingly, the sum of all P (C i ) is equal to one.

With a rectangular probability distribution in one channel, all P (C i) are equal to the reciprocal of the dynamic range. When summing, the dynamic range increases in proportion to the number of summed events (D * n), and the number of states increases in proportion to the dynamic range raised to the power of the number of summed random variables (D ^ n). Such a mismatch arising with increasing n inevitably leads to a change in the law of the probability distribution of random variables, and independently of the initial distribution in each channel, as n grows, it tends to the normal law. Therefore, the first assertion of the PPP was based on summation.

For small values of D and giving N of the order of 10 ⁸ , the probabilities P (C i) are calculated directly in the Microsoft Excel office application. To calculate large values, a simple recursive algorithm was used.

The result of calculating the change in the probability distribution P (C i) when D is 33 depending on n is shown in Figure 2. C (33, 4) is 129, C (33, 8) is 257, and C (33, 16) is 513 ( these three dependencies are located on the figure from top to bottom). The x-axis indicates the relative deviations of the interference from its average most probable value, reduced to a single scale. The greatest probability of all the sums in the above distributions is the sum equal to zero.

According to theory, the probability of at least one of m independent events (independent in probability) is equal to the sum of the probabilities of these events. Adding the probabilities of all sums P (Cξ ± i) to the right and left of the maximum, we obtain the probability of finding the deviation of the noise in the corresponding limits. From figure 2 it can be seen that with a probability of 0.5, the deviation of the interference depending on n changes less significantly relative to the maximum amplitude than for probabilities equal to 0.95 and especially 0.99. With a probability of 0.99, the interference with n equal to 4 is within 73% of its maximum deviation, with n equal to 8 respectively within 53% and with n equal to 16 within 38% of its maximum value. This means that with a four-fold increase in the number of summarized receive channels (the number of independent random variables), the relative amplitude of the interference after summation decreases by about a factor of two.

Using methods of numerical simulation, the obtained approximation was confirmed for large values of the number of channels on a limited interval of the number of samples. The calculation results are summarized in tables 1 and 2. In both tables, the second row shows the calculated value of the SVZ noise signal in channel 1 (SVZ _{sh / channel} ), averaged over the number of channels. Next, the SVZ noise was calculated after summing the n channels, and the ratio of the SVZ _{w / channel} to the SVZ of this sum was determined. It is indicated in the 3rd row of the tables. The 5 lines indicate the error between the values from the 3rd and 4th lines, due to the finite duration of the noise and harmonic signal and confirming the proposed approximation. The difference in results in the tables is clear from their headings. In table 1, the sample value was specified by a random number, and in table 2 by a sinusoid with a random initial amplitude from the D / 2 range of 1024, a random initial phase (φ (x) from the range 0 ÷ 2π) and a random frequency from the KB range - radio waves ), equal to 3–30 MHz, the signal samples were taken with a sampling frequency of 480 MHz (16 samples per period of maximum frequency). The weak difference observed here allows us to proceed to the generation of noise in a given frequency spectrum and shows the validity of the transition from the field of probability theory mathematics to the field of vibration physics, as well as the well-known statement that the spectrum of the sum of signals is equal to the sum of their spectra.

Table 1 f (x, n) = RAND () * 1024-RAND () * 1024 n 128 64 32 16 8 SVZsh / channel 341.29 341, 31 341.32 341.19 341.14 Attitude 11.58 8.19 5.79 4.06 2.90

$$\mathrm{one}/\sqrt[2]{\mathrm{<figure-callout\; id="2"\; label="n"\; filenames="00000028.png,00000029.png"\; state="\{\{state\}\}">n</figure-callout>}}$$

11.31 8.00 5.66 4.00 2.83 Error 2.36% 2.38% 2.31% 1.49% 2.40%

table 2 f (x, n) = [RAND (x) * 1024] * SIN (2 ∗ PI * f (x) + φ (x)) n 128 64 32 16 8 SVZsh / channel 324.49 323.81 322.04 327.53 336.82 Attitude 10.99 7.78 5.59 3.85 2.78

$$\mathrm{one}/\sqrt[2]{\mathrm{<figure-callout\; id="2"\; label="n"\; filenames="00000028.png,00000029.png"\; state="\{\{state\}\}">n</figure-callout>}}$$

11.31 8.00 5.66 4.00 2.83 Error 2.89% 2.70% 1.26% 3.77% 1.83%

The PZP phenomenon occurs not only in the analogues of the invention described above, but also in natural conditions of interference of various sets of signals (waves), the PZP significantly complements the resulting picture of signal addition as a result of interference. At a popular level, it can be interpreted very simply. If, when summing several numbers, some of them are positive and others are negative, then the modulo result will always be less than in the case when all terms are of the same sign. The components of the useful signal as a result of synchronization before summation always turn out to be of the same sign, due to which the signal-to-noise ratio increases.

The above explanation of the PPP shows that the summation and preliminary synchronization of the signal noted in the claims are necessary and sufficient conditions for carrying out the invention, but do not exclude the use of other operations in the processing algorithm. Figure 1a) shows the selectivity characteristics for three groups of channels: 37 channels, 19 channels (from the same 37) and 9 channels from 37 (3, 5, 7, 9, 13, 17, 25, 31, 37), which are very not uniform. The summation of 37 channels does not give a significant improvement in linearity, but the operation of selecting the minimum value from 3 results gives a noticeable improvement in the selectivity characteristic. The number and channel numbers of the antenna field (AP), as well as their coordinates, were calculated taking into account the radial structure of the antenna field. In the center - channel number 1, in the radius R from it - six channels numbered clockwise, then in the radius 2R - 12 channels, then similarly 18.

Turning to a description of the conditions of implementation or the modes under which the proposed solution is implemented, we should dwell on the conditions of independence of random variables, or events that are independent in probability, which should be samples of signals and their sums. To derive the Limitations 1 below for claims 1, 2, the above-described model of the antenna field of the radial structure was used, on which the conditions for the independence of random variables and the spatial selectivity characteristics depending on them were studied. The SVZ changes of a simple harmonic signal at the system output were determined when the direction of reception and the geometric dimensions of the antenna field changed. Numerical modeling was carried out in two dimensions, i.e. the signal source and reception points were located in the same plane. The results of the calculations are shown on the next page in tables 3 and 4.

Table 3 L between n / city 0 42 90 132 180 222 270 312 360 max 60 min. 9.95 9 32.19 25.18 11.46 4.83 5.17 4.62 11.46 23.34 32.19 19 32.19 23.79 7.44 0.08 1.86 0.49 7.44 21.65 32.19 37 32.19 23.53 6.84 0.69 2.71 1.26 6.84 21.38 32.19 132 22 9 32.18 8.96 2.03 8.97 14.13 6.74 2.03 6.58 32.18 19 32.18 4.13 1.25 3.44 6.08 4.63 1.25 0.63 32.18 37 32.17 3.39 0.45 2.81 1.39 2.68 0.45 0.31 32.17

L between n / city 0 42 90 132 180 222 270 312 360 264 44 9 32.13 5.24 17.39 4.35 16.05 8.03 17.39 10.56 32.13 19 32.09 2.04 8.97 3.96 10.76 5.55 8.97 5.00 32.09 37 32.11 0.18 0.68 1.96 5.99 4.96 0.68 2,39 32.11 540 90 9 31.26 12.16 4.80 9.48 5.92 13,99 4.80 12,13 31.26 19 30.89 4.63 4.78 4.46 18.29 3,55 4.88 4.79 30.89 37 30.87 1,52 4.81 1.32 13.75 0.31 4.86 1.20 30.87 λiz. = 100 60 32.19 24.17 8.58 1.86 3.24 2.12 8.58 22.12 32.19 132 32.18 5.49 1.24 5.07 7.20 4.68 1.24 2,51 32.18 264 32.11 2.49 9.01 3.42 10.94 6.18 9.01 5.98 32.11 540 31.01 6.10 4.80 5.09 12.65 5.95 4.85 6.04 31.01

Table 4 L between n / city 0 42 90 132 180 222 270 312 360 max 14,4 min. 2,4 9 2.17 1.78 0.96 0.52 0.49 0.50 0.96 1.68 2.17 19 2.17 1,69 0.71 0.18 0.09 0.16 0.71 1,58 2.17 37 2.17 1.68 0.68 0.15 0.04 0.11 0.68 1,56 2.17 36 6 9 2.17 0.57 0.26 0.52 0.92 0.35 0.26 0.44 2.17 19 2.17 0.19 0.14 0.27 0.41 0.38 0.14 0,07 2.17 37 2.17 0.16 0,07 0.14 0.01 0.11 0,07 0.08 2.17 72 12 9 2.17 0.44 1.00 0.22 0.74 0.55 1.00 0.83 2.17 19 2.17 0.19 0.52 0.27 0.53 0.28 0.52 0.40 2.17 37 2.17 0,07 0.12 0.19 0.23 0.42 0.12 0.21 2.17 144 24 9 2.17 1.06 0.34 1.05 0.68 1.21 0.34 0.84 2.17 19 2.16 0.44 0.16 0.46 0.99 0.56 0.16 0.41 2.16 37 2.16 0.10 0.10 0.02 0.86 0.15 0.10 0.05 2.16 λ = 26 fourteen 2.17 1.72 0.78 0.28 0.21 0.26 0.78 1,61 2.17 36 2.17 0.30 0.15 0.31 0.44 0.28 0.15 0.20 2.17 72 2.17 0.23 0.55 0.23 0.50 0.41 0.55 0.48 2.17 144 2.16 0.53 0.20 0.51 0.84 0.64 0.20 0.43 2.16

The diagrams obtained from them (constructed from the last four columns, where the arithmetic mean values are taken for graphs with n equal to 37, 19, and 9) are shown in figure 3. The figure in table 4 is similar to figure 3 and is not given in the description.

From the above tables and figure 3 it can be seen that with a decrease in the maximum size of the antenna field (L max / min is the distance between two randomly taken emitters) to and less than the emitter wavelength, the steepness of the selectivity characteristic significantly worsens. Obviously, continuing to reduce the size of the field to zero, we get a system equivalent to a single receiver, in which there will be no spatial selection. Here, a phenomenon similar to a physical phenomenon is observed, showing how a wave goes around an obstacle. When the size of the obstacle is the same or smaller than the wavelength, the shadow region does not form, that is, the obstacle bends completely around.

Figures 4a) and 4b) show the selectivity characteristics for three antenna fields (three in one: 37, 19, 9), which were taken with a smaller step of changing the maximum size of the AP for the five detuning angles (14.02; 11.32; 7, 78; 5.17 and 2.58 degrees) between the direction of reception and the direction to the emitter and at a radiator wavelength (λrad.) Of 25. Figure 4a) shows that the configuration of the antenna field and the number of receivers with equal external dimensions have a slight effect on selectivity. The selectivity graphs for n equal to 37, 19, and 9 almost coincide. From figure 4b) it is seen that selectivity begins to appear when the external dimensions of the AP (Lmax.) Begin to significantly exceed the wavelength of the emitter.

When receiving noise distributed in the frequency band, the condition under which they will be attenuated depends on the frequency with the longest wavelength in accordance with expression (9).

According to paragraphs 3, 4 of the claims, the restriction (9) on the operation of the PPP is not imposed with a random nature of the interference in the reception band. If there is a regular component in the interference spectrum, the compensation method of its suppression should be used to increase the signal / noise level.

To demonstrate the validity of Proposition 1, which delivers the basis of the PPP, a simple computational model was created in the Microsoft Excel office application. The first line was indicated by the numbers 16 columns of the spreadsheet. Each cell of all columns starting from the second row to the last with the number 65536 was filled with the expression "= ROUND (RANDIS () * 32; 0)", which generated random integers in them in the range from 0 to 32. The value 32 was chosen, so as in the calculation of the probabilities P (C i) using the recurrence algorithm with a precision of 15 significant digits in Excele, the maximum value of D is 32 for n equal to 16. Then the value obtained in each cell of the resulting table was subjected to the condition “= IF ( A2 = 0; $ AI $ 1; A2) ”, which excluded zero from the set of chi ate to values 32, causing the resulting distribution of random numbers to a strictly uniform law. The obtained number of positive outcomes for each of the numbers from 1 to 32, taken over three columns 1, 8, and 16, is shown in figure 5a). As can be seen from the figure, the distribution is very close to rectangular. After summing over eight and sixteen columns, the resulting distribution was determined, which, as in the case before the summation, is shown as the number of positive outcomes, rather than probability values. To calculate the distribution of numbers we use the PivotTable tool from the MS Excel Data menu.

For understanding and correct interpretation of the graph constructed in figure 5b), we turn to table 5. The number of sums calculated by the expression "(8) and the graph of the number of positive outcomes in fig.5b) are given on a single scale. Since the minimum of the sums dropped out for n, equal to 8, turned out to be equal to 33 in the presented implementation ⁵ ( ⁵ By the implementation of a random number or their set is understood, as is accepted in the literature, the specific value of the dropped random number or their set obtained in MS Excel by recalculation by pressing the "F9" key.) then shk The la starts with this value.The average amounts indicated in the second row of the table turn out to be the most probable and drop out the maximum number of times. tables, as we observe in the above graphs.

Table 5 n = 16 n = 8 Number of Amounts 497 249 Avg the amount 249 125 The number of combinations n of N 1,2089E + 24 1,09951E + 12 Probability cf. amounts 0.01070004 0.014986945 Number of implementations 65534 65534 The number of drops average. amounts 701.23 982.17

Thus, from the graphs in FIGS. 5a) and b) we see that when summing up, the distribution law of random numbers changes, the parameters of which (distribution), in turn, depend on the value of n. Here, one can observe the interference damping mechanism with an increase in the number of summed channels, if we turn to table 6.

Table 6 n = 16 n = 8 min 125 35 Max. 370 215 Number of Amounts 497 249 Deviation limits in% of max. 49.3% 72.3%

с погрешностью, равной -3,7%.The minimum and maximum deviation values close to those obtained in the summary tables and observed visually on the graph of FIG. 5b) are indicated in lines 1 and 2 of the table. Comparing the visible limits in which the interference is located with a probability very close to unity, with the maximum possible deviation according to the expression “Deviation limits = (max-min.) / Number of sums”, we obtain the values in the last row of Table 6. In the previous section, when describing the essence of the invention, it was already noted that with an increase of n by a factor of 4, the interference decays by a factor of 2. For the given implementation of the set of random numbers, the values for n equal to 16 and 8 differ from the square root of 16/8 and equal $$\sqrt[2]{2}$$

with an error of -3.7%.

The validity of Proposition 2 logically follows from the fact that the space-time coordinates of random numbers do not participate in the summation. This fact is checked on the same VM. But here we summarize the samples obtained conditionally as if sequentially in time, taken in one of the columns. For n equal to 8 in the i-th line, the sum includes random numbers X _i + X _{i + 1} + X _{i + 2} + ... + X _{i + 6} + X _{i + 7} obtained in the next eight lines starting from the i-th one. Accordingly, for n equal to 16 in the i-th line, the sum includes random numbers X _i + X _{i + 1} + X _{i + 2} + ... + X _{i + 14} + X _{i + 15} obtained in the next sixteen lines starting from the i-th . The resulting graphs do not differ from those shown in figure 5b).

The validity of Proposition 2 simplifies the construction of a VM for studying the work of the PPP under Proposition 3 with a different (as an example, triangular) initial distribution of independent random variables, as well as the proof of Proposition 4 when calculating the dependence of the relative deviation of interference to a value of n equal to 1024.

To obtain another (triangular) initial distribution of random variables in the first column of the spreadsheet sheet, we use a random function of the form f (x) = (1 + RAND () - RAND ()) * 16, which gives a change in random numbers within the same range. The constructed graphs of the distribution of positive outcomes for him are similar to figure 5b). Data similar to those presented in table 6 for a rectangular distribution are summarized in table 7, from which attenuation at the same rate is also viewed, confirming Claim 3.

Table 7 n = 16 n = 8 min 163 59 Max. 350 197 Number of Amounts 497 249 Deviation limits in% of max. 37.6% 55.4%

Based on Statement 2 of the PPP, we construct a function graph showing the noise attenuation rate as a function of n. To this end, sequentially in each row of the array: we obtain the rectangular distribution over the previous VM, apply accumulation summation, divide by n (getting the average noise deviation), subtract the mathematical expectation and divide by the maximum deviation (which is 16 in our VM). By averaging over 4096 implementations using a simple macro, we get a “damping curve” - an empirical graph of the normalized deviation of the interference from its most probable state, shown for an interval of 1 ÷ 300, in figure 6, which is combined with the maximum error of less than 1.5% with the function

For another (triangular) initial probability distribution, the graph of the named dependence does not visually differ from that shown in Figure 6, with the exception of the starting point, which in the limit is 1/3. Accordingly, the “attenuation curve” for this probability distribution is approximated with an error of less than 1% by an expression similar to (10), but instead of two in the denominator there will be three. Figure 6 is a visual confirmation of Proposition 4.

Figure 6 also shows a graph of the useful signal "Signal 16", obtained by expression (4) as a result of applying the PPP by dividing by n and the maximum deviation of the interference, and which is a straight horizontal line parallel to the axis of change of n. Its distance from the X axis corresponds to the signal level in one of the channels, and the intersection point with two asymptotes corresponds to a n value of 16, above which the signal becomes more noise. With n equal to 1, the noise level is 4 times higher than the signal level, and with n equal to 230 the signal will exceed noise by 3 times.

The graph of figure 6 reflects two approaches: on the left, the claimed solution based on the PPP, where two methods of combating interference are combined (increasing the transmitter power and decreasing the level of interference) and the damping slope is maximum — on the right, the trivial “accumulation method” based solely on the “law big numbers. "

The accumulation method already in its verbal definition contains an indication of the temporal nature of summation, there is no hint of spatial selection, and moreover, the restriction imposed by inequality (9) is not introduced. The proposed method according to claim 1 of the formula makes it easy to achieve many times the improvement of S / N, regardless of the waveform (modulation method). With the accumulation method, which naturally gives some gain, the proposed solution has a very distant similarity. In the accumulation method, the concept of signal synchronization or matching of the signal with the accumulation period is clearly absent, only summation is common.

BRIEF DESCRIPTION OF THE FIGURES

Figure 1a) shows the selectivity characteristics obtained by numerical simulation for three antenna field configurations. The x-axis shows the offset angle between the direction to the radiation source and the direction to the receiving point, the step for calculation was 6 degrees. On average, signal attenuation is 7 times (≈17 dB). The first configuration of the AM includes all 37 receivers, the second - 19 channels of 37 (1 plus all even numbers) and the third - 9 channels of 37 (3, 5, 7, 9, 13, 17, 25, 31, 37). Already at the first deviation of 6 degrees, the total signal fell to the average of the minimum level values. Figure 1b) shows the selectivity characteristic obtained by calculating the smallest value of the three characteristics shown in figa).

Figure 2 illustrates Claim 1 underlying the PPP. The x-axis indicates the relative deviations of the interference from its average most probable value, reduced to a single scale. It can be seen from the figure that with an increase in the probability of interference being within specified limits with increasing n, its relative amplitude decreases. With a probability of 0.99, interference with n equal to 4 is within 73% of its maximum value (deviation), with n equal to 8, respectively, within 53%, and with n equal to 16, within 38% of its maximum value. This means that with an increase in the number of summarized receive channels - (the number of independent random variables) by four times, the relative amplitude of the interference after summation decreases by about half.

The figure 3 shows the characteristics of spatial selectivity for a wavelength of the emitter of 100 meters, from which it can be seen that with a decrease in the maximum size of the antenna field to and less than the wavelength of the emitter, the steepness of the selectivity characteristic significantly worsens. Obviously, continuing to reduce the size of the field to zero, we get a system equivalent to one - the receiver, in which there will be no spatial selection.

Figures 4a) and b) show the dependence of selectivity for three AP configurations with a smooth change in the pitch of the maximum AP size for the five detuning angles (14.02; 11.32; 7.78; 5.17 and 2.58 degrees) between the direction of reception and direction to the emitter and when the emitter wavelength (λrad.) is 25. Figure 4a) shows that the configuration of the antenna field and the number of receivers with equal external dimensions have little effect on selectivity. The selectivity graphs for AP configurations with the number of emitters equal to 37, 19, and 9 almost coincide. From figure 7b) it is seen that the selectivity begins to appear when the external dimensions of the AP (Lmax.) Begin to significantly exceed the wavelength of the emitter.

From the graphs in figure 5a) (the distribution of three channels out of 16 for spatial sampling before summation is shown) and 5b) (distribution after summation) we see how, in accordance with Proposition 1 of the PPP, the distribution law of random numbers changes, the parameters of which are given for two values of n . With an increase in the number of summed channels, the visible limits in which the interference is located with a probability very close to unity decrease by 1.4 times. For the given implementation of the set of random numbers, the values of this deviation of the interference from its most probable value for n equal to 16 and 8 differ from the square root of two with an error of -3.7%.

Figure 6 shows the "attenuation curve" - an empirical graph of the normalized deviation of the interference from its most probable state for a rectangular initial probability distribution. The graph is shown for an interval of 1 ÷ 300, which, with a maximum error of less than 1.5%, is combined with the function 1 / (2 × ROOT (n)). The “Signal16” line also shows the normalized signal level, which, when crossed with asymptotes for 16, gives a S / N level equal to one.

The figure 7 shows the waveforms of the input and output signals of the system obtained using the computational model shown in the previous figure. Figure 7a), from top to bottom, shows the time variation of the input signal received from the received source, the resulting noise field signal and their sum in one of the 37 receiving channels.

In the upper graph of Fig. 7b), after the synchronization described previously, the resulting samples of the total noise signal are reflected when the received signal is turned off. The lower graph of figure 7b) shows the output signal of a system that uses the PPP according to claim 1 of the formula, and the combined graph of the input signal without noise in one of the channels, amplified 10 times.

The figure 8 presents the waveform of the input and output signals. Figure 8a) reflects the signals in one of the 37 receiving channels, from which it can be seen that the signal and interference are approximately equal, but not synchronized. Modern means when using one receiver do not allow in real time to separate the signal and interference lying in the same band, with this ratio. At the output of the system, Figure 8b), you can see the ratio of the total output signal and interference (in the absence of a received signal), as well as good synchronization of the envelopes of the input and output signals. To avoid overlapping waveforms of the total signal and the received signal (in the absence of interference), different amplifications were applied to them.

Figure 9a) and b) show the spectra of the AM signal and Narrowband interference obtained using the discrete Fourier transform (DFT) for the number of samples equal to 2048. Figure 9a) in the 40 kHz bandwidth, which accounts for only 9 out of 2048 basic harmonics of the DFT, the narrow-band nature of the modeled interference is traced.

The harmonic components of the interference spectrum that are out of the passband amount to approximately 10% of the total value of all harmonics. On the contrary, the spectrum of the AM signal shown in Figure 9a), the side frequencies of which coincide with the base harmonics by selection, does not have side harmonics. The spectrum of the system output signal obtained in a system of 37 receivers is shown in FIG. 9b). The level of out-of-band harmonics and noise residues does not exceed 2% of the output signal-to-noise mixture taken in the reception band.

The table in figure 10 shows that the broadband coefficient for a standard glass multimode optical fiber and a wavelength of 1300 nm is 600 MHz / km. The corresponding values in columns J, N, and P, are marked in gray, clearly indicate this pattern. It can also be seen from the results given in the table that the propagation of signals in the fiber does not occur contrary to, but due to the intermode dispersion and the operation of the BCP. F (delay), calculated as the reciprocal of the delay between the fastest and slowest modes, is compared with the actual bandwidth in the last row of the table. Due to the work of the PZP, this ratio increases with increasing length of the fiber. The chart below shows the input signal, and the bottom chart shows the output received in the mode indicated in column N of the table. The light line on it marks the threshold (row of the table "Level. Ogr."), By which it is possible to reliably decode the bit sequence. The second, fourth and penultimate pulses are single. The data in the table columns reflect the “islands of passage” of the bit sequence; for other ratios of the number of modes and the length of the fiber, there was no passage (the possibility of visual decoding).

In figure 11 there are two diagrams. The upper diagram displays the values of the samples to be decoded (dark bars) and single pulses (light bars) of the 63-bit source code sequence, the values of which are multiplied by the arithmetic mean of the samples to be decoded. Here, the applied decoding algorithm is clearly visible, the same for all four computational models according to claim 3 of the formula. If the estimated sample is less than the arithmetic mean level, then its bit is assigned the value zero, and vice versa. In the case of an error in the diagram during the modeling process, a discrepancy is observed when at least one source single bit exceeds the level of the decoded sample, or a knowingly zero sample exceeds the average level. The diagram shows a situation with no errors.

The bottom diagram is a graph of the five characteristics from table 12, which shows the gain when applying the PZP for the option when one random UTF process is superimposed on all k times repeating bands of the amplitude-manipulated signal. Since for every k harmonics of the spectrum of the signal in this VM there is only one harmonic of the soft starter, the poor synchronization necessary to obtain a gain when applying the PZP, due to the fact that the periods of the carrier frequencies in each of the bands are different, does not lead to a lack of gain. It can be seen from the table and graph that even with four parallel transmission bands of the same bit sequence in real time, the S / N at the input becomes less than one. And with 16 repetitions, it reaches minus 7.44 dB.

Figure 12 shows fragments of the waveforms of the four VM processes shown for two time scales (more at the top and less on the bottom diagram) with a value of k equal to 16, when the band of the transmitted signal is repeated k times, and the band of the soft starter is expanded in proportion to the 1st harmonic of a single pulse . From top to bottom it shows: noise (soft starter, black), a signal mixture with noise (dark gray), a modulated signal without noise (light gray) and the original bit sequence (white line). Due to the overlapping of the process oscillograms on top of the oscillogram, the signal plus noise mixture graph is placed in the upper layer for better visualization, and for reference, a table of values of the corresponding bits is shown below the original bit sequence. Without this table, on oscillograms over a long period of processes, it is difficult to determine what the sequences of various radio pulses correspond to. Thanks to the a priori known parameters of the transmitted signal received against the background of strong interference, it is possible to sufficiently reliably obtain all the transmitted information from it due to the use of the PPP.

Figure 13 shows the waveforms of the initial portion of the signal samples for two values of k obtained on the VM for table 4. The light line shows the mixture of signal and noise before summing the repeating segments, the dark line shows the straightened mixture after summing. Both signals, input and output, correspond to the same SCP implementation. The upper graph shows the signal transmission mode at k equal to 16, the lower one shows the “transmission failure” at k equal to 15. The upper graph clearly shows that the second unit pulse at the output corresponds to a dip in the input mixture, and the third and fourth significant excess of the input pulses ( and with a shift of fronts) over the weekend. Fronts and slopes of single output pulses are quite clear, such a sequence is easy to decode. On the lower graph, against the background of the input mixture, practically neither the signal itself, nor single pulses are visible, which in the VM was manifested in a catastrophically increased number of decoding errors. Obviously, there is nothing to decode here.

The figure 14 shows two diagrams of the decoding processes according to claim 4 of the claims. It shows the simultaneous implementation of -1 and 3 claims that are not physically separable, since fragments of the signal in the mixture with interference are taken to obtain a time delay (repetition) from different points in space, and the propagation medium plays the role of a delay line. The decoding of the bit sequence was carried out in two ways, the traditional described in previous models and the method of moving summation, when each time sample (its weight) at the output of the RTS is the sum of the n-th number of previous and n-th number of subsequent samples.

The figure shows an example of decoding the first 32 of the 63 bits of the pulse sequence with a k value of 16. The upper diagram against 32,000 input samples shows the output (heavier) samples in bright tone, and the dark line decoding threshold that matches the average value for all input samples input mixture signal and interference. If this threshold is exceeded, we have a logical unit at the output of the RTS and vice versa. The lower diagram shows for clarity, combined on a timeline with the upper diagram, the values at the input of the previously described decoder and the original bit sequence, the amplitude of the single pulses of which is equal to the decoding threshold.

DETAILED DESCRIPTION OF THE INVENTION

For a visual representation of the feasibility and capabilities of the claimed method according to paragraph 1 of the claims, a computational model is created that shows the possibility of extracting a signal from a noise background when its power level (SVZ) exceeds the signal level. In this model, a noise signal is generated by generating harmonic signals with a random amplitude in a limited dynamic range, a random frequency from the shortwave range of the radio waves, and a random initial phase within 360 degrees.

The algorithm of the computational model consists of two parts. In the first part (1) in a conditionally specified region of space (for simplicity of calculations on the plane) a noise field is formed and then, by summing with accumulation, a dense spectrum is created from simple harmonics in a given frequency band and with a given power (2). Similarly, by setting the coordinates and amplitude, the modulated signal to be received is created (3). According to the simple law of the propagation of a spherical wave in space without taking into account various losses, in addition to the spatial decrease in power, the propagation of interfering harmonics and signal to the receiving sites is simulated (4). Then, in the generated AP, the values of the total signal at the reception points (5) are calculated. As a result of the application of control elements in the model, the noise power without a signal, the signal without noise and the total signal, as well as the signal-to-noise ratio (6), are separately calculated.

The computational process was performed in the following sequence:

- the received signal is turned off;

- using a simple macro, the values of the samples are summed with accumulation from 1024 harmonics;

- after calculating the SVZ values for the noise in 37 reception channels, the SVZ value of the received signal is selected so that the signal-to-noise ratio at the system input (i.e., in each of the 37 reception channels) is less than zero decibels;

- when summing both the signal and the interference, the reception was synchronized from the point where the emitter of the received signal was;

- after which, the summation of the samples (mixing) of the signal and interference was carried out and the total summation of the samples of 37 channels, the measurement of the SVZ and signal-to-noise ratios at the system output.

- Simultaneously with the formation of an array of signal and interference samples, a spectrum of input harmonics was formed in one of the channels.

The results are shown in figure 7 and in table 8.

Table 8 ENTRANCE SV3s / SV3sh Db Max Sign / Noise 1,0082616 0,0715 Min Sign / Noise 0.4548466 -6.8427 Wed meaning Sign / Noise 0.6664117 % rel. Sign + Noise -3.5251 EXIT SV3s / SV3sh Noise Summ37 0,0051957 27.84% Sign / Noise Signm Summ37 0.0152464 81.69% Db Sign + Noise Summ37 0.0186639 100.00% 9,3505

Figure 7a) shows the top-down change in time of the input signal received from the received source, the resulting noise field signal and their sum in one of the 37 reception channels.

In the upper graph of Fig. 7b), after the synchronization described previously, the resulting samples of the total noise signal are reflected when the received signal is turned off. The lower graph of figure 7b) shows the output signal of a system that uses the PPP according to claim 1 of the formula, and the graph of the input signal combined with it without noise from one channel amplified by 10 times.

From both figures a) and b) it can be seen that at the input of any of the 37 channels, the received signal is destroyed by interference, the strict periodicity of the total signal disappears. While at the output of a multi-channel system we see an output signal that is visually closely synchronized with the input signal received, and to some extent modulated in amplitude and phase by an interference signal.

From table 8 it is also seen that at the input of a multichannel system that uses spatial selection due to the operation of the PZP, the signal / noise level is less than unity, and at the output, a significant gain is obtained by the ratio of the levels of SVZ. Touching the same values expressed in decibels, we can talk about negative input signal-to-noise values relative to the described reception system.

The second example, confirming the method according to claim 1, is based on a similar computational model that uses the spatial selection algorithm, but differs from the previous one in the way the noise signal is generated at the input of the system. Here, with some additional restrictions, we use the model of narrow-band fluctuation noise, described in [16], and which is given by the formula:

u _p (t) = U _pm (t) SIN [ω ₀ t + φ _p (t)],

where U _pm (t) and φ _p (t) are the random amplitude and phase of the narrow-band fluctuation process.

The introduced restrictions are related to the discrete nature of the signal and consist in limiting the limits of variation of the random amplitude and phase during the transition from the current noise sample to the next one. The values obtained in the received amplitude-modulated signal are taken beyond the limits of such changes. To limit the limits of variation of the random amplitude, a fluctuation process model was used, which differs from the above formula and is based on the expression for the amplitude-modulated signal, but, in addition to the random phase in square brackets, it contains a random phase at the modulating frequency.

The received narrowband interference (SCP) signal is formed by introducing two random phases into the AM signal according to the expression:

Where U _am (N) - N-th sample of the amplitude-modulated signal

A is the amplitude of the signal, f _n is the carrier (central) frequency of the signal;

F _t is the sampling frequency of the signal; Ω is the modulating frequency of the signal;

N = 0, 1, 2 ...; - signal sample number; m = 0.5 - modulation index of a narrowband interference sample U _upp (N) are formed recurrently in terms of:

φ _a0 (ψ) = 2 * [ψ (01) -0.5] * k * π; - the initial phase of the first sample changes in the amplitude of the interference;

φ _aN (ψ) = 2 * [ψ (01) -0.5] * j * φ (Ω) + φ _{a (N-1)} (ψ); - random phase of the Nth sample of the change in the amplitude of the interference;

A _N (ψ) = A * {1-m * SIN [2 * π * N * Ω / F _t + φ _aN (ψ)]}; - random amplitude of the Nth interference sample:

φ _ξ0 (ψ) = 2 * [ψ (01) -0.5] * k * π; - the initial phase of the first sample phase change of the carrier frequency of the interference;

φ _ξN (ψ) = 2 * [ψ (01) -0.5] * i * ξ (Ω) + φ _{ξ (N-1)} (ψ); - random phase of the N-th sample of the phase change of the carrier frequency of the interference;

Where - φ (Ω) = ξ (Ω) = 4 * π * Ω / F _t ; - the maximum phase change from sample to sample to obtain a random amplitude and phase of the carrier frequency; j, i = 1, 2, 3, 4.

ψ (01) = 0 ÷ 1; - a random number with a uniform distribution ranging from zero to one. Thanks to the expression 2 * [Ψ (01) -0.5], the phase changes in the positive or negative direction. The PPP starts to work already at k greater than or equal to 1/10.

In the process of generating the received signal (11) and narrow-band interference (12), two parameters were monitored - SVZ and the maximum change between adjacent samples, which for the signal and interference did not differ from one implementation to another by no more than 10-12%.

The applied computational model used one source of the received signal and one source of narrow-band interference with the coefficients J and i equal to unity, A is 1024, F _f is 20480000 Hz, f _n is 3040000 Hz and Ω is 40,000 Hz. The characteristic of spatial selectivity was measured depending on the angle between the signal transmission signal and the signal interference, the latter moved around the circle around the antenna, as well as the characteristic confirming the restriction 1 in expression (9) when changing the overall dimensions of the antenna. The measurement results are shown in Table 9 for the first characteristic and in Table 10 for the second.

Table 9 angle (deg.) one 2 3 four 5 6 7 Wed Value Vh. Sign / Pom. (dB) -1.918 -1.842 -1.953 -1.988 -1.885 -1.952 -1.921 Out Signal / Interference (dB) -1,339 -0.878 -0.743 0.156 1,774 3,812 5,624 distance PRD signal. - PRD pom, (m) 1745 3490 5235 6980 8724 10467 12210 angle (deg.) 8 9 10 eleven - 90 180 Wed Value Vh. Sign / Pom. (dB) -1.967 -1.911 -2.005 -1.876 -1.867 -1.970 Out Signal / Interference (dB) 8,113 12,800 15,365 17,943 21,574 2,913 distance PRD signal. - PRD room (m) 13951 15692 17431 19169 141421 200,000

Table 10 The angle between the signal Tx and the Tx interference is 10 deg. step 45 40 35 thirty 25 twenty fifteen 10 5 2 max AP size (m) 540 480 420 360 300 240 180 120 60 24 Out Sign / Pom. (Db) 15.46 11.24 8.04 4.91 2.77 1.19 0.19 -0.61 -1.36 -1.58

Table 9 shows that, starting from 4 degrees, the signal-to-noise ratio at the system output becomes positive with a negative value at the input. A decrease in gain at 180 degrees indicates the presence of a back lobe in the radiation pattern of the system.

According to table 10, it can be seen that for AP sizes comparable with the wavelength (about 100 m), the gain in the system disappears (step 10).

Oscillograms of the input signals and output are presented in figures 8a) and b).

Since the signals from a distance of 100 km underwent some attenuation according to the law described above for the first example, for a visual assessment of their relative values at the input and output of the system, they were amplified with various coefficients indicated in the figures. On figa), reflecting the signals in one of the 37 reception channels, it can be seen that the signal and interference are approximately equal, but not synchronized. Modern means when using a single receiver do not allow in real time to separate the signal and interference with such a ratio. At the output of the system, figb), you can see the ratio of the total output signal and interference in the absence of a received signal, as well as good synchronization of the envelopes of the input and output signals.

Figure 9a) and b) show the spectra of the AM signal and Narrowband interference obtained using the discrete Fourier transform (DFT) for the number of samples equal to 2048. For clarity, the coefficients J and i are taken equal to 4. The values of the spectra were calculated over an interval of 240 harmonic components . In figure 9a), in the 40 kHz bandwidth, which accounts for only 9 of the 2048 DFT harmonics, the narrow-band nature of the modeled interference is traced.

The harmonic components of the interference spectrum that are outside the passband amount to approximately 10% of the total value of all harmonics, and their occurrence is due to the mismatch of the instantaneous harmonic components of the noise with the basic frequencies of the DFT (in the range of two hundred and forty). On the contrary, the spectrum of the AM signal shown in Figure 9a), the side frequencies of which coincide with the base harmonics by selection, does not have side harmonics. The values of its harmonics, as can be seen from the figure, are between the values of the components of narrow-band noise averaged over 16 realizations and the maximum.

The spectrum of the system output signal obtained in the system of 37 receivers described in this example is depicted in FIG. 9b). The level of out-of-band harmonics, noise residues, calculated according to the method described in the previous paragraph, does not exceed 2% of the output signal-to-noise mixture taken in the reception band.

Thus, in two examples, confirming the possibility of carrying out the invention according to claim 1 of the formula, it is seen that the task of receiving signals weaker than noise is feasible and requires knowledge of the direction to the radiation source.

To analyze the operation of the PPP according to claim 2 of the formula in the field of optics, based on the data obtained from [17], the next computational model was created. A computational experiment to verify the transmission of digital information via a fiber-optic channel confirmed the validity of a number of statements obtained, probably, empirically, and revealed the operation of a PCP in the process of receiving a bit sequence at the output of a fiber. The forty-eight bit sequence of the word “Signal” - “D1 E8 E3 ED E0 EV” was applied to the input of a glass fiber with a core diameter of 50 nm; source wavelength 1300 nm (in vacuum), core loss 0.8 dB / km. The calculations were carried out according to formulas (3) - (8) given in the article by V. Yakovlev, an employee of the St. Petersburg company PROSOFT. By changing the difference in the refractive indices of the shell and the core during the simulation, the required number of modes was selected, the fiber length L (m) was changed, and by changing the duration of one bit of information, the passband P (props). The criterion for these changes was the possibility of visual decoding of the bit sequence when comparing it with the input diagram.

The experimental results are shown in the table in figure 10. They coincide quantitatively and qualitatively with the statement from the article that the broadband coefficient for a standard glass multimode optical fiber and a wavelength of 1300 nm is 600 MHz / km. In this case, with an increase / decrease in the distance by half, the band decreases / also increases by half. In the table, the corresponding values in the columns that clearly indicate this pattern are marked in gray (columns J, N, P).

It can also be seen from the results given in the table that the distant propagation of signals in the fiber does not occur contrary to, but due to intermode dispersion. Due to intermode dispersion, together with the phenomenon of a change in the phase of the wave upon reflection by 180 degrees, as well as the automatic summation of the modes at the output of the fiber, the distance is. which is capable of transmitting information at a given speed increases significantly. The modes, passing a different path, have, in addition to different intensities, different phases. As a result of summation (the work of the PZP), some of the modes in antiphase cancel each other out and neutralize the negative effect resulting from dispersion. The selection of the necessary modes at the output of the fiber for today is a difficult technical task. In addition, in the process of movement, due to various bends of the fiber, they constantly move from the fast category to the slow category and vice versa. The probability distribution here is two-dimensional.

The calculations found that because of the time delay between the fastest and slowest mode, which (delay) limits the bandwidth of the output signal, the resulting bandwidth is much less than real. The parameter P (delay) in the 5th row of the table, calculated as the reciprocal of the above propagation delay of various modes, is compared with the real bandwidth in the last row of the table. The ratio of the real band, which is obtained due to the operation of the PZP, to the band limited by the signal delay, increases with increasing length of the fiber.

The diagram below the table of figure 10 shows the input signal, and the bottom diagram shows the output signal obtained in the mode indicated in column N of the table. The light line on it marks the threshold (line “Level.Org.” Of the table), by which it is possible to reliably decode the bit sequence. The second, fourth and penultimate pulses are single.

Another example that confirms the principle of the PPP according to claim 1 of the formula and Restriction 1 is the arrangement of the organs of vision in the vast majority of fauna. The size of the eyes relative to the wavelength of the received light satisfies condition (9), and the presence of two eyes is associated with a low level of spatial selectivity of the multichannel system in the radial (along the beam) direction.

The implementation of the invention according to claim 2 is possible with a limited number of receivers, which is obvious from the VM operation algorithm for the case of wide and narrowband interference, but if inequality (9) is mandatory for AL size (example: parabolic antenna, directional microphone).

The application of the PPP according to claim 3 of the claims for one receiver fits into the well-known laws of information theory, which describe the transmission of signals through communication channels. It is known that the volume of the signal Vc transmitted through the noise communication channel is equal to the product of the throughput C and the transmission duration Tc and for the case with white noise according to the formula of K. Shannon [12] it is:

where F _s , P _s , P _W - band, signal power and noise, respectively.

From (13) we obtain the expression for the exchange of communication channel parameters when changing the noise environment (P _c / P _W ).

Where h ² is the ratio of P _with / P _W ; in the initial (input) noise environment, aq ² equal to P _s / P _w ; accordingly, with a changed (output) noise environment.

If the S / N level worsens, the time-frequency parameters of the communication channel should be changed. Considering the radio engineering system (RTS) as a communication channel, which can be both passive and active with respect to gain, the reverse transformation is performed for active RTS, increasing the S / N ratio to a level above critical, at which the decoder provides reliable information transmitted source. For the gain-generating radio system, the calculated equality for the exchange of the band and signal duration for the S / N ratio has the same form (14). In such a RTS, the parameter h ² should be considered output (decoder input), and q ^{2 the} input S / N ratio. From (14), three methods (options) arise for obtaining a gain with respect to S / N when transmitting the same signal volume from the input to the output of the RTS. A gain can be obtained by changing (increasing) either the transmission time, or the signal band, or both at the same time. In the first version of (14) we will have:

where i is Input, o is output.

For calculations, k is taken equal to T _o / T _i from a number of 1, 2, 3 ... n; and accordingly, the gain received in the RTS will be determined from the expression:

The longer the transmission time, the greater the theoretically possible gain obtained in the communication system.

By analogy in the second embodiment, we also take k taken equal to F _o / F _i from a number of 1, 2, 3 ... n; and we get the same expressions (15) and (16). Despite the similarity of the expressions for the first two options, when exchanging the transmission time to the S / N level in order to obtain a gain Z, the transmission speed changes (falls), in the second embodiment, the gain is achieved in real time. Extracting the square root from both sides of expression (15), we obtain, by analogy with (16), the dependence of the gain z (zet is small) relative to the value of the SVZ taken at k equal to 1, which is obtained by changing k and is denoted below in the figures as “Shannon curve ".

The initial expression for calculating the gain in the exchange of the band and the duration of the transfer to the S / N ratio according to the third embodiment will take the form:

where E _w and F _w, respectively, the spectral density and the noise band that change together with the band at a constant signal power, ak ₁ and k ₂ are coefficients showing the rate of change of the transmission time and the band occupied by the signal, respectively.

For the third option, when both the transmission band and the time change, there is a practically significant and quite simple realizable special case when, with an increase in the signal bandwidth Fc, the transmission time Tc decreases by the same amount. At the same time, it becomes possible to repeat the signal to bring the information exchange process to real-time mode and thereby obtain a certain gain in the signal-to-noise ratio. For this particular case, according to the third option, a computational model was developed.

The validity of expressions (14) and (17) follows both for a hypothetical theoretically possible case, since it is a consequence of the law and / or a condition for preserving the transmitted information, and for a practical case verified on a computational model (VM). Since both the right and left equal signs reflect the same amount of information, decoded at the optimal decoding threshold. For the same purpose, a modulation method — amplitude manipulation — was chosen in the VM.

On the four simplest computational models for a harmonic signal with amplitude manipulation, with the help of which a 63-bit sequence was transmitted, the winnings obtained by using the PPP for the above cases were determined.

- In the first embodiment, an increase in the transmission time of Te was achieved by simply repeating in time k times the original signal. The narrowband interference model (SCP) used in the computational model has been described above.

- According to the second option, two cases were modeled. In the first case, an increase in the band of the transmitted signal was made by repeating the band of the original signal in the adjacent higher frequency band due to a change in the carrier frequency. In this case, the SCP band expanded simultaneously with an increase in Fc by a factor of k by increasing the modulating frequency, and the carrier (central) noise frequency was shifted to the center of the signal band. Based on the assumption that the spectral density of the interference power is constant, it in this VM was corrected by the formula:

N = P ₁ / F _l = P ₂ / F ₂ = P ₂ / (F1 * k); => P ₂ = P ₁ * k; whence SVZ ₂ = ROOT (k) * SVZ ₁ ;

The presence of a gain was also checked for the second case, in which the noise band increased k times with the signal by repetition (as a signal in the first case). Moreover, to obtain a gain similar to the first case, it was required to straighten the signal + noise separately in each of the k-th bands. In practice, this means having k receive channels.

- In the third embodiment, an increase in the Fc band by a factor of k and a decrease in Tc by a factor of k was achieved by reducing by k times the duration of a single pulse of an information bit. Shortening the length of a single pulse means increasing k times the frequency of its first harmonic, and, consequently, the band occupied by the signal. The SCP band was varied in proportion to the value of the first harmonic of single bit pulses. Then, to bring the signals to real-time mode, the 63-bit sequence was repeated k times. The spectral density of the interference power here was also corrected by the formula given for the second option.

For all three variants, the output S / N ratio (for k = 1), at which, according to the information theory [9], [12], a stable decoding of the bit sequence is performed, was chosen and naturally turned out to be h ² ≈10 dB. The dependences of the theoretically limiting “Shannon curve” and the curve obtained at the four indicated VMs when applying the PPP according to claim 3 are given in tables 11, 12, 13, 14. The decoding threshold (PD) value shown in the figures was selected empirically according to the level of 15-30% of the average value at the output of the adder, when the number of decoding errors of zeros and ones over 16 implementations of soft starters, over which the values of the gain z as a function of z (k) were averaged, was reduced to zero. When the AP is less than 15-20%, the number of errors in the 16 implementation of the SCP exceeds 2, 3 cases. And with AP greater than 30%, the value of the gain z decreased. The presence of a small number (≤3) of errors shown in the tables in the figures indicates the correct choice of the decoding threshold, that is, the optimality of the obtained PD.

Table 11 number of repetitions Decoding threshold S / N input (SVZ) S / N input (dB) S / N output (SVZ) Sum of errors Out / In ratio Shannon Curve The win relates. one 27% 3.24 10,2 3.24 3 one one one 2 33% 2.75 8.80 3.70 0 1.34 2.15 1.18 3 35% 2,36 7.47 4.04 one 1.71 2.98 1.37 four 34% 2.07 6.31 4.15 one 2.01 3.65 1,57 5 35% 1.87 5.42 4.14 one 2.22 4.23 1.74 6 33% 1.70 4.62 4.33 0 2.54 4.74 1.90 7 35% 1,53 3.72 4.49 one 2.93 5.21 2.11 8 31% 1.38 2.78 3.82 2 2.78 5.63 2,36 9 31% 1.23 1.79 4.02 2 3.27 6.03 2.64 10 thirty% 1.13 1,07 3.98 one 3.52 6.41 2.87 eleven 33% 1,07 0.60 3.88 0 3.62 6.76 3.03 12 25% 1,03 0.27 3.94 2 3.82 7.10 3.14 13 34% 1.01 0.08 4.07 0 4.03 7.42 3.21 fourteen 37% 0.98 -0.14 4.04 0 4.11 7.73 3.30 fifteen 29% 0.96 -0.32 3.77 0 3.91 8.02 3.36 16 34% 0.95 -0.48 4.12 0 4.36 8.31 3.43 -0.5 dB

Table 12 number of repetitions Decoding Threshold S / N input (SVZ) S / N input (SVZ) Sum of errors Spectrum. tight powerful N Attitude Shannon Curve The win relates. one 43% 3.04 3.00 2 6.99 0.98 1.00 1.00 2 twenty% 2.05 2.61 2 6.96 1.27 2.15 1.48 3 24% 1.27 1.82 one 6.99 1.44 2.98 2.40 four 17% 0.90 1.39 3 6.99 1.55 3.65 3.39 5 16% 0.76 1.25 one 6.97 1,64 4.23 4.00 6 19% 0.71 1,57 3 7.01 2.20 4.74 4.27 7 17% 0.72 1.74 3 6.99 2.43 5.21 4.25 8 19% 0.71 1.75 one 6.99 2.48 5.63 4.31 9 19% 0.69 1.84 one 6.97 2.66 6.03 4.41 10 22% 0.62 1.81 2 6.98 2.91 6.41 4.89 eleven 21% 0.58 1.71 one 6.97 2.96 6.76 5.26 12 twenty% 0.54 1,57 one 6.99 2.93 7.10 5.68 13 eighteen% 0.50 1.47 2 6.99 2.91 7.42 6.04 fourteen 21% 0.47 1.48 2 6.98 3.12 7.73 6.44 fifteen twenty% 0.45 1.41 2 6.98 3.17 8.02 6.82 16 twenty% 0.42 1.34 one 6.97 3.15 8.31 7.17 -7.44 dB

Table 13 number of repetitions Decoding Threshold S / N input (SVZ) S / N output (SVZ) Sum of errors Attitude Shannon Curve The win relates. The win relates. On ex. one 44% 3.09 3.09 2 1.00 1.00 1.00 1.00 2 40% 2.54 2,39 0 0.94 2.15 1.22 1.30 3 42% 2.25 1.92 0 0.86 2.98 1.38 1,61 four 41% 2.02 1.66 0 0.82 3.65 1,54 1.86 5 37% 1.84 1.42 0 0.77 4.23 1,69 2.18 6 34% 1,69 1.27 one 0.76 4.74 1.84 2.43 7 33% 1,58 1.14 0 0.72 5.21 1.96 2.72 8 34% 1.49 1,02 2 0.69 5.63 2.08 3.04 9 29% 1.41 0.95 one 0.67 6.03 2.20 3.26 10 28% 1.33 0.86 one 0.65 6.41 2,32 3,59 eleven 28% 1.27 0.79 0 0.62 6.76 2.44 3.91 12 24% 1.21 0.74 2 0.61 7.10 2,56 4.19 13 25% 1.16 0.68 one 0.59 7.42 2.67 4,53 fourteen 24% 1.12 0.66 0 0.59 7.73 2.77 4.68 fifteen eighteen% 1,08 0.63 one 0.58 8.02 2.86 4.91 16 eighteen% 1,04 0.59 3 0.57 8.31 2.97 5.25 0.36 dB -4.59 dB Table 14 Repeat Number
one Number of repetitions 2 Decoding Threshold S / N input (SVZ) P Signal relates. Sum of errors Attitude Shannon Curve The win relates. one one 38% 1.92 1.00 0 0.99 1.00 1.00 2 2 38% 1.66 1.22 one 1.15 2.15 1.16 3 3 29% 1,50 1.35 one 1.27 2.98 1.28 four four 31% 1.40 1.46 0 1.36 3.65 1.37 5 5 35% 1.36 1.55 3 1.32 4.23 1.41 6 6 36% 1.30 1.66 one 1.47 4.74 1.48 7 7 33% 1.25 1.72 0 1,53 5.21 1,54 8 8 34% 1.17 1.73 one 1,62 5.63 1,63 9 9 31% 1.14 1.78 one 1.68 6.03 1,69 10 10 27% 1.10 1.79 3 1.43 6.41 1.74 eleven eleven 29% 1.06 1.82 one 1.46 6.76 1.81 12 12 31% 1,04 1.87 3 1.84 7.10 1.85 13 13 37% 1,02 1.92 one 1.87 7.42 1.88 fourteen fourteen 37% 1.01 1.96 0 1.90 7.73 1.91 fifteen fifteen 27% 0.97 1.95 one 1,61 8.02 1.98 16 16 thirty% 0.94 1.95 2 2.04 8.31 2.05 -0.57 dB

Decoding was carried out by summing pre-taken modulo samples along the entire length of a single pulse. The synchronization of the initial phase of the summation process was carried out, as an example on the model for one of the options, using a simple correlator in which the initial phase was selected according to two criteria: by the minimum of PD and the minimum of decoding errors.

The AP values in the tables represent the minimum distance between the smallest single sample and the highest zero sample of the entire 63-bit sequence as a percentage of the average level. The S / N output level, as the ratio of the SVZ signal to the SVZ noise, was determined in all VMs before decoding. Accordingly, the “Ratio”, as one of the criteria for the gain, is the ratio between S / N output to input. The “relative gain” was calculated similarly to the “Shannon curve” as the ratio S / N input at k equal to 1 to the current value. “The sum of errors” is the sum of errors over 16 SCP implementations, over which all measured characteristics of computational models were averaged, i.e. the number of errors out of 1008 bits. All simulation results were obtained with j and i equal to 1 in the soft starter settings.

It can be seen from the tables that when applying the PPP, in all cases, the gain in the S / N ratio steadily increases with increasing k, but it turns out to be less than the theoretically limiting one (as in the case of white noise according to the K. Shannon formula). They also show that after a certain value of k, the S / N ratio at the input of the RTS becomes less than unity. Therefore, for claim 3 of the claims, the use of the PPP allows to achieve a negative (in dB) input S / N ratio.

The obtained results of calculations according to the first embodiment are shown for illustration purposes in FIG. 11, in which two diagrams are located. The upper diagram displays the values of the samples to be decoded (dark bars) and single pulses (light bars) of the 63-bit source code sequence, the values of which are multiplied by the arithmetic mean of the samples to be decoded. Using the latter, the decoding algorithm used is clearly visible, which is the same for all four computational models. If the estimated sample is less than the arithmetic mean level, then its bit is assigned the value zero. And vice versa, if it is more than the average level, then - one. In the event of an error in the diagram, a discrepancy is observed when at least one source single bit exceeds the level of the decoded sample, or a knowingly zero sample exceeds the average level. The diagram shows a situation with no errors.

The bottom diagram is a graph of the five characteristics from table 12, which shows the gain when applying the PPP for the first case according to the second option, when one random SCP process is superimposed on all k times repeating bands of the amplitude-manipulated signal. Since for every k harmonics of the spectrum of the signal in this VM there is only one harmonic of the soft starter, the poor synchronization necessary to obtain a gain when applying the PZP, due to the fact that the periods of the carrier frequencies in each of the bands are different, does not lead to a lack of gain. The table and graph show that even with four parallel transmission bands of the same bit sequence in real time (unlike the first option), the S / N input (SVZ) becomes less than one. And with 16 repetitions, it reaches minus 7.44 dB. The power spectral density of the soft starter remains almost constant as a result of the correction described above, and the curve of "relative gain" reaches a value of more than seven.

Figure 12 shows fragments of the waveforms of the four VM processes according to Table 12, shown for two time scales (more at the top and less at the bottom diagram) with a k value of 16. From top to bottom it shows: noise (soft starter, black), signal mixture with noise (dark gray shade), a modulated signal without noise (light gray shade), and the original bit sequence (white line). Due to the overlapping of processes on each other in the upper waveform, the graph of the C + W mixture is placed in the upper layer for clarity, and for reference, a table of values of the corresponding bits is shown below the original bit sequence. Without this table, on oscillograms over a long period of processes, it is difficult to determine what the sequences of various radio pulses correspond to. Thanks to the a priori known parameters of the transmitted signal received against the background of strong interference, it is possible to sufficiently reliably obtain all the transmitted information from it due to the use of the PPP.

Due to the unsatisfactory process of sampling synchronization (the periods of the carrier frequencies in each of the bands are different) in the third VM, without preliminary rectification of the C + N mixture in each channel for the second case according to the second option (table 13), it was not possible to manage to get a tangible gain. Here, for each harmonic of the spectrum of the signal, in contrast to the previous VM, there is one harmonic of the SCP spectrum, therefore, as a result of summing before decoding, not only the noise but also the signal is attenuated. This VM provides an example of synchronization of the useful signal along the envelope, since carrier-frequency synchronization with spectral transfer for addition (application of the PZP) would significantly complicate the VM. Despite the fact that the S / N at the input barely reaches unity, it is obvious that in the continuation it will inevitably cross the zero (in dB) value. In contrast to the VM for the first case, the gain obtained at the output of the decoder here reaches minus 4.59 dB.

The results of obtaining a win for the particular case noted above in the third embodiment are shown in Table 14. The complexity of the synchronization process, as an integral condition for carrying out the invention, was most clearly revealed in this VM. Bringing the transmission and reception modes to real-time by repeating the signal, the required number of times was performed in two ways, which are reflected in two rows of values "number of repeats 1 and" number of repeats 2 "marked with gray color" failures in the passage "of signals. A similar phenomenon, described above for the case of information transmission through the fiber, was observed due to the mismatch of the phases of different modes, where “islands of transmission" appeared. In this computational model, against the general positive background of growth in gain, as a result of an increase in the number of repetitions, “passing failures” were discovered.

A different arrangement of the dips was obtained due to the fact that to obtain the signal lifetime, as for k equal to 1, with all repetitions for the sequence “number of repetitions 1”, a certain error was allowed in the calculations, while for the second sequence “the number of repetitions 2 »An additional rounding of the intermediate result with correction of the signal lifetime due to a change in the sampling frequency was introduced into the calculation process. For those values of k at which dips occurred, the passage could not be achieved even by an increase of an order of magnitude in the amplitude of the input signal. The lack of transmission was manifested in a catastrophically increased number of decoding errors. For these values of k, the reduced values of the characteristics indicated in the table and in the graphs were obtained by interpolation. The values of the characteristics marked by the passage for both sequences coincided with great accuracy.

Figure 13 shows the waveforms of the initial portion of the signal samples for two values of k. The light line shows the mixture of signal and interference before summing the repeating segments, the dark line shows the rectified mixture after summing. Both signals, input and output, correspond to the same SCP implementation. The upper graph shows the signal transmission mode at k equal to 16, the lower one shows the “transmission failure” at k equal to 15. The upper graph clearly shows that the second single pulse at the output corresponds to a dip in the input mixture, and the third and fourth significant excess of the input pulses (and with a shift of fronts) over the weekend. Fronts and slopes of single output pulses are quite clear, such a sequence is easy to decode. On the lower graph, against the background of the input mixture, practically neither the signal itself, nor individual pulses are visible. Decoding is excluded here.

Taking into account the fact that the frequency band does not belong to the time category, but rather is a spatial characteristic, the second option with the expansion of the signal band can also be attributed to claim 1. Accordingly, the third option, when the gain is obtained by expanding the band and by increasing / decreasing the transmission time, should be referred to claims 1 and 3 at the same time.

The next VM according to claim 4 is constructed for the degenerate case when only one bit of information is repeated k times and the narrowing of the band occurs due to an increase in the length of a single pulse. The input S / N ratio was selected equal to 3 dB at k equal to 16, at an ultrasonic frequency of 36.7 kHz.

It shows the simultaneous implementation of 1 and 3 claims that are not physically separable, since fragments of the signal in the mixture with interference are taken to obtain a time delay (repetition) from different points in space, and the propagation medium plays the role of a delay line.

The decoding of the bit sequence was carried out in two ways, the traditional described in previous models and the method of moving summation, when each time sample (its weight) at the output of the RTS is the sum of the n-th number of previous and n-th number of subsequent samples. For a sound speed of 330 m / s and 895 summed samples, the total length of the chain of receivers located on one line in the direction of the emitter was 1.14 meters. The example shown in this VM can actually be used to transmit telemetric information or remote control commands in the ultrasonic range [18]. The data presented in the table show how, with a k-fold change in the length of a single pulse, the number of errors at the decoder output decreases asymptotically, and figure 14 shows two diagrams of the decoding processes.

number of repetitions one 2 3 four 5 6 7 8 9 10 eleven 12 13 fourteen fifteen 16 number of errors 161 91 45 44 37 24 12 7 eleven four  2 3 four one 0 0

When the number of repetitions changes, the SCP band, its amplitude, and the signal amplitude did not change. The figure also shows an example of decoding the first 32 of the 63 bits of the pulse sequence with a k value of 16. The upper diagram against 32,000 input samples shows the output (heavier) samples in bright tone, and the dark line decoding threshold that matches the average value for all input samples of the input mixture signal and interference. If this threshold is exceeded, we have a logical unit at the output of the RTS and vice versa. The lower diagram shows for clarity, combined on a timeline with the upper diagram, the values at the input of the previously described decoder and the original bit sequence, the amplitude of the single pulses of which is equal to the decoding threshold. It can be seen from the upper diagram that the values of zeros and ones clearly do not follow from the input sequence of samples, and after rolling summation, the bit sequence, regardless of the information transfer rate, is quite reliable.

If it is necessary to increase the S / N during the transmission and reception of signals with angular modulation, the application of the PZP can be performed in two approaches. In the first case (according to claim 3 of the formula), it is first necessary to perform decoding, that is, convert the angular modulation to the amplitude modulation, and then apply envelope synchronization and summation in accordance with the PPP. Angular modulation decoding can be performed, as an example, by the expression:

Where U (t) - The value of the current envelope reference;

K is the coefficient of proportionality;

Ui, Ui-1 - values of two adjacent samples;

Umax - the maximum value of the signal samples in the demodulation interval.

sin ^-1 is the arcsine function.

In the second case, when using the option with expanding the signal band to obtain a gain in the S / N ratio, perform synchronization on the carrier frequency with the transfer of the signal spectra into a single band for summing. Then perform demodulation. According to claim 1 of the formula, the type of modulation does not matter.

BIBLIOGRAPHY

No. Lit. Literature, link on the Internet. Page one. Diversity reception systems (SRP) for the Altai Center, page 1, paragraph 2. one = 1.2_3 & 1 = Rus 2. Phased array antenna. one 3. Optimal diversity coherent and incoherent reception in channels with fluctuation and lumped interference. A.A. Sikarev. “Problems of Information Transmission”, Volume IX, Issue 1. 1973, p. 58, para. 2. 2 .mathnet.ru / php / getFT.phtml? irnid = ppi & paperid = 882 & what = fullt & option_lang = rus On the issue of improving the organization of intermittent mobile communications with diversity reception of signals in a channel with lognormal fading. M. Adrianov. Journal “Technologies and means of communication” No. 2, 2009, p. 1, para. 7. http://www.tssonline.ru/articles2/fix-op/k-voprosu-o-povvshenii four. A method of spatial diversity receiving a signal from a radiation source transmitted over a multipath channel, and a device for its implementation. Figovsky E.A .; Bakharev O.D. RF patent No. 2075832, 03/20/1997, see the formula. 2 5. Gutkin L.S. The theory of optimal methods of radio reception with fluctuation interference, M.-L., 1961; Kharkevich A.A., Fighting interference, M., 1963. 2 , p. 2, para. 4. 6. Radio Magazine, No. 8, 1999. V. POLYAKOV. Synchronous AM receiver. Page 31. 3 ftp: //.ru/pub/arhiv/. 7. RADIO Magazine No. 2 of 1991, A. RUDNEV. Medium-wave receiver with a synchronous detector. Page 56 3 ftp: //ftp.7/ftp.radio.ru/r%3eub/arhiv/. 8. Radio wave propagation Black F.B. 2nd ed., Ext. and recycling. - M. “Sov. Radio ”, 1972, p. 71. 3 ftp: //ftp.kiaml1.http: //rssi.ru/pub/gps/book/black/.

9. Security algorithm. Number 3 2004 V. Belkin. Radio channel notification system. Antenna feeder path. Page 16, paragraph 9. 3 = 12189 && pos = 1 & stp = 10. 10. The Lyapunov Central Limit Convergence Theorem and his theorem “The Law of Large Numbers” are described. 7 http://www.exponenta.ru/educat/class/courses/tv/theme0/10.asp. eleven. The Laplace integral theorem on the variation of the dispersion of a sum is proportional to the root of the number of terms. 7 _ver / 6_1 /. 12. Electronic textbook on Information Theory. 3.5. Continuous Channel Bandwidth. Shannon's theorem. Expressions (3.55), (3.53) and comments on them. 7, 30, 33 13. Signals and linear systems Topic 2. Space and metrology of signals. 2008 Davydov A.V. (tss02-Signal metrology.doc; last paragraph, expression (2.2.1)). 7.8 fourteen. T. Korn and G. Korn. Math reference book for scientists and engineers. Ed. "Science", M. 1978, p. 580. 8 fifteen. T. Korn and G. Korn. Math reference book for scientists and engineers. Ed. "Science", M. 1978, pp. 550, 573. 8 16. K.Yu. Agranovsky, D.N. Zlatogursky, V.G. Kiselev; Radio engineering systems. M. Higher School, 1979, p. 60, expression (2.1). 25 17. Modern automation technology No. 4 of 2002, p. 74. V. Yakovlev. The basics of fiber optic technology. Page 74.
http: // 29th eighteen. Great Soviet Encyclopedia / Ultrasound. Page 1, paragraph 2. 39

Claims (7)

1. A method of increasing the signal-to-noise level (ratio) during spatial selection of signals, comprising the operation of synchronizing and then summing, characterized in that all components of the same signal source are received at spatially separated points, the mutual spatial coordinates of which must be in the region, the maximum dimensions of which are much greater than the maximum wavelength of the received wavelength range, and the extraction of the signal from the mixture with noise is successful by choosing the required number of terms achieved even when in each of the received channels the noise level significantly exceeds the level of the useful signal. 2. The method according to claim 1, characterized in that the synchronization of the components of the same signal is carried out by means of a time delay. 3. The method according to claim 1, characterized in that the step of receiving components of the same signal contains a limited number of receiving points. 4. A method of increasing the signal-to-noise level (ratio) when changing the time-frequency parameters of the signal, comprising the operation of synchronizing the time-frequency elements of the input signal and then summing, characterized in that all time-frequency elements of the input signal are formed on the transmitting side, and for which the time-frequency parameters of the realization of the noise received with the signal in the same band must satisfy the independence conditions for random variables, and extracting the signal from the mixture with noise successfully by choosing the required number of terms is achieved even in the case when the noise level in the input signal significantly exceeds the level of each time-frequency element of the useful signal. 5. The method according to claim 4, characterized in that the synchronization of the time-frequency elements when they are separated in time is carried out by means of a time delay, the role of which is played by the propagation medium. 6. The method according to claim 4, characterized in that noise is artificially introduced into the signal structure on the transmitting side, which exceeds the useful signal in level. 7. The method according to claim 1 or 4, characterized in that, depending on the signal-to-noise level, a reduced number of spatial or time-frequency terms are adaptively selected for summing.

Images