RU2502077C1 - Method of estimating signal-to-noise ratio over section of harmonic oscillation interval - Google Patents

Abstract

FIELD: radio engineering, communication.

SUBSTANCE: disclosed method is based on separate estimation of noise power and the power of a signal and noise mixture. Noise power is calculated by comparing a received signal and a filtered signal and squaring the difference thereof and the power of the signal and noise mixture is calculated by squaring the received signal. The method provides continuous payload transmission.

EFFECT: obtaining an estimate of the signal-to-noise ratio over a unit interval, having more accurate characteristics under the effect of various types of interference over the unit interval without introducing redundancy.

1 dwg

Description

The invention relates to the field of telecommunications and can be used in data transmission systems, when operating at a given speed in a single-frequency mode without introducing redundancy, to evaluate the quality of a communication channel.

In modern information transfer systems using multi-parameter adaptation methods, the problem arises of prompt and accurate estimation of the signal-to-noise ratio (SIR) in the process of message transmission. OSB data on each elementary package is also necessary to implement soft decoding procedures. Therefore, the task of determining the OSB on each elementary premise is relevant.

Currently, there are known methods for estimating the SIR using primary signal parameters, such as amplitude, phase, distortion of the edges of the packets, etc. [one]. A significant drawback of these methods is the need to obtain a sufficiently large volume of sample values, which does not allow forming an OSB estimate for each elementary basis on their basis. However, in HF channels, due to the presence of fading, the magnitude of the SIR on each elementary package is different.

A device for measuring the signal-to-noise ratio is described in RF patent No. 2332676, which allows filtering to separate the power of the noise component from the power of the signal-noise mixture, generating signals proportional to the signal and noise power. Thus, the signal-to-noise ratio is obtained.

The disadvantage of this method is that the signal-to-noise ratio is set for a certain type of interference and does not take into account possible changes in signal and noise parameters, thereby reducing the measurement accuracy. If the interference power in the signal band will differ from the interference power outside the band, then the accuracy of this method drops sharply.

The closest way to the claimed one is the method of measuring the signal-to-noise ratio, described in RF patent No. 2117954, which allows, through the accumulation and correlation processing of phase information, to detect a useful signal against a background of noise (interference) and isolate the values of the phase fluctuation function due to the influence of interference. Based on the obtained dependences, the variance of phase fluctuations is analyzed and the signal-to-noise ratio is determined. The measured signal-to-noise ratio can be used to assess the reliability and accuracy of radar information or the quality of communication channels.

The signal-to-noise ratio measured in this way is obtained by processing a sufficiently long input signal, since the phase fluctuation function is actually a sample of phase values obtained for each signal pulse (signal sending). Thus, it is not possible to obtain the signal-to-noise ratio value for a single pulse (sending).

The aim of the present invention is to provide a signal-to-noise estimate for the duration of an elementary packet, which has higher accuracy characteristics when exposed to various types of interference, when the data transmission system operates in the mode when a single-frequency signal is applied on the duration of one elementary parcel, without introducing redundancy.

This goal is achieved by detecting the useful signal, filtering the signal, comparing the received and filtered signal and analyzing their differences based on the obtained dependence of the estimated power of the noise component on the true power of the noise component, and then calculating the signal-to-noise ratio based on the known dependence of the received signal power on the power noise components.

The structural diagram of the proposed method is shown in figure 1.

The problem is solved as follows.

If a segment of harmonic oscillation is used as a signal, then, regardless of the type of modulation and the initial phase, it is possible to represent the signal in the form of the following model:

S (t) = А · cos (ωt + φ),

where A is the signal amplitude, ω is the circular frequency, φ is the initial phase, t is time.

Then the received signal for the duration of the elementary symbol at the input of the receiver can be represented as follows:

,

,

where B is the signal amplitude, ψ is the initial phase, which changed as a result of the influence of the communication channel, ξ (t) is the noise.

In the discrete case:

X _k = B cos (ωΔtk + ψ),

ξ _k = ξ (Δtk),

,

,

where ω is the signal frequency, Δt is the interval between two adjacent samples, k is the reference number, and ψ is the initial phase.

It is known that equidistant samples of harmonic oscillations of frequency ω are interconnected by the following recurrence relation:

X _k = 2 cos (ωΔt) X _k-1 -X _k-2 .

We use this relation to predict the estimate of the value of the subsequent signal sample from the two previous samples of the signal received from the channel:

.

.

,

. Сформировав выборку этих величин объемом N, выражение для оценки среднеквадратичного отклонения измеренного отсчета от прогнозируемого определяется выражениемThus, at each kth step, using this predicted filter, you can get a pair

,

. Having formed a sample of these quantities with volume N, the expression for estimating the standard deviation of the measured reference from the predicted one is determined by the expression

и

совпадают, поэтому оценка

является несмещенной.It should be noted that all quantities included in the expression are available for measurement and calculation based on the measured values. Since expectation

and

match therefore the score

is unbiased.

The mean square deviation of the predicted value from the measured value is also determined as follows:

- это дисперсия шума при измерении, то естьWhere

is the noise variance in the measurement, i.e.

.

.

If the noise characteristics do not change during signal processing, then

In this case, the expression for the mean square deviation of the predicted value from the measured value can be written as follows:

It is known that σ ² = P _W , that is, it is the power of the noise component. Therefore, the power rating of the noise component is determined by the following relationship:

The estimate of the power of the signal-to-noise mixture is determined by the expression

For cases h ² >> 1 a fairly accurate estimate is the value:

It should be noted that this condition in most cases is satisfied with the stable operation of real communication systems.

The described method works as follows.

The received signal is fed to an analog-to-digital converter 1, in which signal samples are obtained, which then enter the accumulation unit 2, where they are accumulated over a time interval corresponding to the duration of the elementary package. Further, from the accumulation unit 2, the accumulated array of samples is fed simultaneously to the multiplier 3, the predictive filter 6 and the multiplier 7. In the multiplier 3, the obtained array of samples is element-wise multiplied by itself, that is, it is squared. The resulting new array is fed to the adder 4, in which all elements of the array are summed, and the result is transmitted to the divider 5, in which it is divided by the length of the array. In the predictive filter 6, the array of samples is filtered and the filtering result is transmitted to the adder 8. In the multiplier 7, the array is multiplied by minus one and the resulting array is passed to the adder 8. In the adder 8, the arrays coming from the predictive filter 6 and the multiplier 7 are summed the result is passed to the multiplier 9, in which the element array is multiplied by itself, and the result is passed to the adder 10. In the adder 10, the elements of the resulting array are summed and transmitted the result is a divisor 11, in which the obtained value is divided by the coefficient K = (4 · cos ² (ωΔt) +2) · N, where ω is the signal frequency, Δt is the interval between two adjacent samples, N is the length of the array. The result is transmitted to the divider 12. In the divider 12, the value coming from the divider 5 is divided by the value coming from the divider 11, thus obtaining the desired signal-to-noise ratio estimate.

Literature

1. I.T. Rozhkov. Synthesis of signal-to-noise ratio meters for received radio signals. Saratov University Press, 1991.

Claims (1)

A method for evaluating the signal-to-noise ratio over the duration of an elementary transmission, which can be used in data transmission systems, when operating at a given speed in a single-frequency mode without introducing redundancy, for assessing the quality of a communication channel, characterized in that the signal is received on the receiving side, it is fed to an analog-to-digital converter one, in which signal samples are obtained, which then enter the accumulation unit two, where they are accumulated over a time interval corresponding to m of the duration of the elementary premise, and the formation of the corresponding array of samples, which is simultaneously fed to the multiplier three, the predictive filter six and the multiplier seven, while in the multiplier three the element array of the samples is multiplied by itself, that is, it is squared, and the resulting a new array is fed to the adder four, in which all the elements of the array are summed, and the result is transmitted to a divider five, in which it is divided by the length of the array, while in the predictor the filter six filters the array of samples and passes the filter result to the adder eight, while the multiplier seven multiplies the array by minus one and passes the resulting array to the adder eight, in which the arrays coming from the predictive filter six and the multiplier seven are summed pass nine into the multiplier, in which the resulting array is element-wise multiplied by itself, and the result is passed to the adder ten, in which the elements are summed. th array and transmitting the result to divider eleven, wherein the divided value obtained in advance calculated coefficient K = (4 · cos ² (ωΔt) +2) · N, where ω - frequency signal, Δt - interval between two adjacent samples, N - the length of the array, and the result is also transmitted to the divider twelve, in which the value coming from the divider five is divided by the value coming from the divider eleven, thus obtaining the desired signal-to-noise ratio estimate for the duration of the elementary packet without introducing redundancy.