RU2809969C1 - Method for assessing signal/noise ratio in wireless communication devices - Google Patents

Abstract

FIELD: electrical engineering; communications.

SUBSTANCE: to increase the noise immunity of the communication system, a method is proposed for assessing the signal-to-noise ratio in wireless communication devices, which uses a modified nonparametric rank test based on the Mann-Whitney test. The result of the rank calculation is resistant to significant deviations in the amplitude of the samples of the sampled signal, since the range of values of the estimation functional is limited. In this case, to receive multi-position manipulation signals and form a noise sample, n=2^r + 16 incoherent correlators are used, where r is the number of binary symbols per frequency position, the channel block of incoherent correlators is connected to the decision device block, the decision device processes the received data in the array, for each significant signal frequency position the rank value relative to the noise sample is calculated, the differences between the convolution values of the signal and noise samples are calculated, a general difference sample of the resulting convolution differences is formed, and the modulo value of the convolution differences is ranked.

EFFECT: increasing the noise immunity of the communication system.

1 cl, 3 dwg

Description

The invention relates to the field of radio engineering and communications, and can be used for demodulating multi-position frequency telegraphy signals with pseudo-random tuning of the operating frequency (PRFC), ensuring the functioning of adaptive noise-resistant radio communication lines in the decameter range of 1.5-30 MHz with mobile marine objects (MMO).

Modern software management and automation systems require constant information support on various aspects of their operation under the influence of both unintentional and intentional interference. This leads to the need to increase the noise immunity, throughput and noise immunity of the communication system with the software.

The most promising solution to ensure the fulfillment of these requirements for decameter radio links seems to be the use of algorithms for assessing the quality of the received signal, resistant to deviations from the hypothesis of a Gaussian distribution of interference. Such interference with a non-Gaussian power spectral density distribution can occur under conditions of frequency selective fading and pulsed interference.

In communication complexes with software, in which noise-resistant radio links are implemented, signal quality in the decameter sub-band is assessed by analyzing detected errors in the channel redundant noise-resistant code. There are known methods for assessing changes in signal phase and calculating signal and interference powers, as well as their complex combination in a single algorithm. Methods based on the analysis of detected errors in a channel redundant noise-correcting code have a disadvantage associated with the correcting abilities of the noise-correcting code. Radio links in the decameter range are characterized by the occurrence of frequency-selective fading as a result of the arrival at the receiving point of several rays that have traveled paths of different lengths. The value of the phase difference between two beams fluctuates in time; as a result of the addition of two beams with approximately equal amplitudes and fluctuating phases, the amplitude-frequency and phase-frequency characteristics become periodic (frequency-selective fading), as a result of which error groups are formed in the stream of information symbols. Therefore, to correct and detect them, it is advisable to use cyclic block codes, for example, Reed-Solomon codes. The correction and detection ability of Reed-Solomon codes depends on the redundancy of the code. With a significant increase in the number of check symbols, the information transmission rate in the radio link decreases.

A known device and method for assessing interference and noise in the communication system RU2324291C2 [1]. The product contains a correlation device, a noise and signal separation device, and a reference sequence signal selection device. In this method of estimating the signal-to-noise ratio, calculations of the signal level and interference level are performed using quadratic quantities (calculation of the correlation function for each subcarrier of the OFDM signal). In turn, any discrete sample that differs from the hypothesis of a normal distribution of interference can significantly affect the resulting estimate (integration is carried out by multiplying the amplitudes of the discrete samples of the signal transmitted through the communication channel and the amplitudes of the discrete samples of the reference signal, then summing the resulting products). We can conclude that in difficult conditions of interference and multiplicative signal distortions that occur in multipath channels. This method for assessing interference and noise in a communication system has a disadvantage caused by the instability of the assessment indicators.

The prototype of the proposed method is a method for estimating the signal-to-noise ratio in wireless communication devices with receive diversity RU 2349034C2 [2], which consists in the fact that the estimation module generates estimates of traffic symbols and pilot signal for the received wireless signal based on spatial samples of the signal, scales the estimates traffic and pilot symbols using a spatial projection Wiener filter function, and estimates the signal-to-noise ratio based on the scaled estimates of the traffic and pilot symbols. This method has the disadvantage that the estimation result is unlimited, characteristic of the mean square error, and the limit taken relative to the quadratic estimation functional, with an increase in the number of sample elements, provided that the values of the sample elements are not limited and differ from the main distribution (Tukey model), will diverge [3]:

where ε is the “small” probability of a significantly different value appearing;

- distribution function of a normal random variable;

m _x - mathematical expectation (shift parameter) of a random variable having a normal distribution;

σ _x - standard deviation (scale parameter) of a random variable having a normal distribution;

Δ(x) is the distribution function of an abnormal random variable.

Methods for assessing signal quality based on estimating signal power and interference using the least squares method or based on non-robust correlation methods, in the presence of significant deviations in sample values (outliers) caused by frequency-selective fading, which causes nonlinearity of the phase-frequency characteristic of a multipath channel, or pulsed interference , can give estimates significantly different from optimal ones.

Sampling a mixture of useful information signal and additive noise may differ significantly from the hypothesis of a normal (Gaussian) distribution. To obtain an optimal stable estimate of signal quality under specified conditions, nonparametric statistical tests or robust versions of parametric statistical tests can be used.

Nonparametric are distribution-free statistical tests in which the probability of falsely rejecting the null hypothesis (the sample does not fit the distribution, there is no relationship between samples) is the same for all possible continuous distributions of the main hypothesis.

Rank assessment methods are non-parametric criteria.

In particular, the Mann-Whitney rank estimation method is used to test the hypothesis about the presence of a useful information signal in a signal/noise mixture noisy with non-Gaussian noise, such as impulse noise, and also subject to multiplicative distortion (frequency-selective fading) [4].

Taking into account the identified contradictions caused by the instability of the quadratic estimation functionals to the deviation of the distribution of values from the Gaussian distribution, we can conclude that it is advisable to assess the signal quality using a statistical nonparametric assessment of a sampled analog signal of multi-position frequency telegraphy (MSFT) based on rank tests. This allows you to evaluate the quality of the signal and make a decision on its adaptation in the presence of very different values in the statistical sample. As well as in the case of using a channel quality assessment in a radio link based on the error vector of an error-correcting code (channel level of the EMIOS) obtained during decoding, situations are likely to arise when the correcting abilities of the code will be insufficient and the channel encoder will refuse to decode when the number of errors that can be exceeded is exceeded. be corrected.

The disadvantage of both the analogue and the prototype is that they do not have the ability to make a stable assessment of the signal quality, which does not allow making the optimal decision on adapting the parameters of the radio channel according to the operating frequency and the parameters of the signal-code design. No technical solutions have been found whose set of features coincides with the set of features of the proposed method of demodulation and evaluation of the quality of the received signal based on the modified Mann-Whitney rank test for multi-position frequency shift keying (telegraphy) signals with pseudo-random tuning of the operating frequency, including distinctive features. Obtaining such a set of features is a new technical method that ensures the above-mentioned technical result, which allows us to conclude that the claimed invention meets the “inventive step” criterion.

The purpose of the present invention is to increase the noise immunity, throughput and noise immunity of a communication system with mobile marine objects in conditions of both unintentional and intentional interference.

This goal is achieved due to the fact that the method for assessing the signal-to-noise ratio in wireless communication devices uses a modified non-parametric rank test based on the Mann-Whitney criterion; the result of the rank calculation is resistant to significant deviations in the amplitude of samples of the sampled signal, since the range of values of the evaluation functional limited, the method differs in that to receive multi-position manipulation signals and form a noise (reference) sample, it uses n = 2 ^r + 16 incoherent correlators, where r is the number of binary symbols (bits) per one frequency position, a block of channels of incoherent correlators is connected n outputs to the solver block, through which the results of the operation of convolution of the signal sample and convolution of the noise (reference) sample are transmitted in the form of an array, in the solver the received data is processed in the array, for each significant signal frequency position the rank value relative to the reference noise is determined samples in accordance with the modified Mann-Whitney algorithm, for which the differences in the convolution values of the signal and noise reference samples are calculated, a combined difference sample of the resulting convolution differences is formed, the modulus of the value of the convolution differences is ranked, the rank value is calculated by summing the indices of the corresponding positions of the combined ranked difference sample . Based on the calculation of the rank of each information signal frequency position, a decision is made on the value of r bits at the information output of the decision device, while the signal from the pseudo-random sequence generator is received at the input of the decision device and on its basis the nominal operating frequency is calculated taking into account the selected sub-range in which pseudo-random adjustment of the operating frequency is performed, at each clock cycle of receiving frequency positions, the maximum calculated rank is put in correspondence with the frequency subrange, the obtained maximum values of the rank of signal frequency positions and the calculated value of the nominal operating frequency represent an assessment of the quality of signal passage in the communication channel in the corresponding subrange of operating frequencies, this the estimate in the form of a two-dimensional array is fed to the input of the information and control system of the communication complex in order to provide the functions of frequency and code adaptation of the signal-code design, operational adjustment of frequency schedules compiled using radio wave propagation forecasting programs.

The essence of the invention is illustrated in Fig. 1 and fig. 2.

A method for assessing the signal-to-noise ratio in wireless communication devices, based on demodulation of multi-position frequency shift keying (telegraphy) signals with pseudo-random tuning of the operating frequency and assessment of the quality of the received signal based on the modified non-parametric Mann-Whitney rank test, is presented in Fig. 1, where the following notations are adopted:

1 - pseudo-random sequence generator (PSG);

2 - frequency mixer;

3 - frequency generator controlled by PSP;

4 - analog-to-digital converter (ADC);

5 - block of incoherent correlators;

6 - deciding device;

7 - information and control system.

The PSP generator 1 generates a pseudo-random sequence (with uniform distribution), receives clock pulses from the unified time system (UTS) of the PMO and key information for the synchronous operation of the transmitting and receiving sets of channel-forming equipment. The pseudo-random sequence enters the generator of highly stable high-frequency oscillations, controlled by PSP 3, to form harmonic oscillations with a frequency determined by the PSP value. At the same time, a pseudo-random sequence is fed into the decision device to form the value of the nominal frequency at which the radiation is carried out and the formation of an array of signal transmission quality estimates.

Frequency mixer 2, based on the signal coming from the operating frequency path (preselector) and a highly stable harmonic oscillation from the generator controlled by PSP 3, generates a signal with a constant frequency and then transmits it to the ADC.

The frequency generator, controlled by PSP 3, receives a sequence (random number) from the PSP generator, on the basis of which the nominal frequency of the high-frequency oscillation is formed, and then the frequency is fed to frequency mixer 2.

ADC 4 converts the MRCT signal with a constant frequency into a digital I/O signal (sampled and quantized by quadrature) and outputs it to the block of incoherent correlators.

Block of incoherent correlators 5 carries out convolution of the digital I/O signal. In this case, frequency shifts between frequency positions (subcarriers) ΔF _subcarriers are selected with the condition of ensuring orthogonality in a strong sense between the nominal values of the frequencies of each frequency position.

where Δƒ is the width of the spectrum of a radio pulse with a duration t (each frequency position).

Also, in the block of incoherent correlators, the reference noise sample is calculated from 16 orthogonal in the amplified sense (relative to the information signal) frequency positions. The signal and reference samples are transmitted to the decision device 6 (see Fig. 2).

In the decision device 6, the data of the signal and reference samples, the values of the bandwidth, the value of the correlation interval n, expressed by the number of discrete samples of the information sample during the duration of one radio pulse, are processed. Next, a modified Mann-Whitney rank test is performed, based on the results of which the rank of each significant frequency position is calculated. A decision is made in favor of the significant frequency position of the MRCT signal that received the highest rank value. This same maximum rank value is an assessment of the quality of reception of the MFCT signal at the operating frequency, taking into account the frequency converter. Next, the decision device produces a demodulated information signal to the channel decoder, and a data array formed from the nominal operating frequencies and rank assessment values is transmitted to the information and control system of the complex for adapting the radio link according to the operating frequency and signal-code design.

The information and control system 7 generates, on the basis of estimates received in the decision device, control signals for adapting the radio link and issues them to the demodulator devices, as well as to the transmitting part of the channelization equipment for transmission via the feedback channel. At the same time, a data array is generated with the results of reception at operating frequencies, operating time and ranking assessment, which can be used to perform machine learning with reinforcement, as well as adapt other decameter channels of the communication complex with software.

In fig. Figure 2 shows the software algorithm for the functioning of the decision device, which implements a method for calculating and generating a data array with the results of assessing the signal quality in each frequency subrange.

Let us accept the following notation:

i is the number of the significant information position of the MRCT signal;

t - time variable;

τ is the duration of one radio pulse of the MRCT signal;

n is the number of the discrete sample of the sampled MRCT signal;

rank s _i (nΔt) - rank of the significant i-th information position of the MRCT signal;

N is the number of elements of the information sample;

K is the number of elements of the reference (noise) sample;

M is the number of significant positions of the MRCT signal;

s _i (n) - sampled MRCT signal;

- conjugate (according to Hilbert) sampled MRCT signal;

z _i (n) is a sampled mixture of the MCHT signal and additive white Gaussian noise (the signal can be distorted by the multiplicative effect).

It is advisable to form the signal-code design of the MPCT for decameter range channels with a random phase on the basis of radio pulses orthogonal to each other in an amplified sense. For such a case, the following expression is valid:

at

Then, by calculating the correlation function using an incoherent (quadrature) correlator, we obtain samples for each of the significant frequency positions and a noise (reference) sample:

From the information and control system 7 of the communication complex, the block of incoherent correlators 5 receives data on the frequency shift between the frequency positions of the MFCT signal, expressed in the nominal frequency relative to the nominal intermediate frequency at which the signal enters the ADC, and the time correlation interval, which is expressed by the number of discrete samples digital carrier of the MFCT signal entering the block of incoherent correlators 5.

The samples enter the solver 6, where the Mann-Whitney rank test is performed relative to the reference sample, which includes calculating the matrix of differences between the signal and noise samples D _i,k :

Considering that We present the matrix of differences D _i,k in the form:

The difference matrix D _i,k is transformed into a row matrix R _1,MK by combining the rows and ranking its elements in increasing order of the differences by the value of their modulus (absolute value) in the row matrix R _1,MK :

For each frequency position s _i (nΔt) of the MFCT signal, the rank value 203 is calculated by summing the ordinal indices belonging to the corresponding row of the matrix D _i,k which contains the differences for the corresponding frequency position:

Based on the calculated rank statistical estimates of each frequency position, a decision is made about the symbol received from the communication channel. The maximum rank of a frequency position is an assessment of the quality of signal transmission in the frequency subrange in which the signal was emitted.

To form a data structure of the results of assessing signal reception in the frequency subrange, data from a generator controlled by PSP 3 is received at the input of the decision device, with the help of which the nominal operating frequency 204 is determined in the solver. The data is presented in the form of an array in which the result of the ranking is assigned to the subrange estimates rank s _i (nΔt) 205. To perform the radio channel adaptation function, the resulting array is transferred to the information and control system of the communication complex.

Simulation modeling carried out for the correlation estimation method and the rank estimation method using the modified Mann-Whitney algorithm made it possible to calculate the relative (asymptotic) efficiency of the rank nonparametric test relative to the correlation statistical criterion, presented in Fig. 3.

The calculation results show that if there are values in the sample that differ significantly from the normal distribution, the efficiency of the correlation estimate is 75% lower than the ranking estimate with the same statistical sample sizes:

Thus, the use of the proposed method for assessing the quality of the received signal makes it possible to increase the efficiency and accuracy of the assessment, reduce the adaptation time by making an optimal decision to change the parameters of the signal-code design, the operating frequency range, ensuring an increase in the reliability, throughput and noise immunity of decameter radio communication channels with pseudo-random adjustment of the operating frequency with moving marine objects under the influence of both unintentional and intentional interference.

Literature

1. Patent RU2324291C2 (RF). A device and method for assessing interference and noise in a communication system. / Zae-Hwan Chang (KR) Yun-Sang Park (KR) Sang-Hoon Sung (KK) Chan-Young Kim (KR) - Publ. 04/30/2004 - H04L 1/206.

2. Patent RU 2349034C2 (RF). Estimation of signal-to-noise ratio in wireless communication devices with receive diversity. / Farrokh Abrishamkar (US), Brian Banister (US), etc. - Publ. 05/27/2006 - Н04В 1/1027.

3. Hampel F., Ronchetti E., Rausseu P., Stael W. Robustness in statistics. Influence function approach. - M.: Mir, 1989.

4. Huber P. Robustness in statistics. - M.: Mir, 1984.

Claims (1)

A method for assessing the signal-to-noise ratio in wireless communication devices, using a modified non-parametric rank test based on the Mann-Whitney criterion. The result of the rank calculation is resistant to significant deviations in the amplitude of samples of the sampled signal, since the range of values of the evaluation functional is limited, the method is characterized by the fact that for n=2 ^r +16 incoherent correlators are used for receiving multi-position manipulation signals and forming a noise sample, where r is the number of binary symbols per one frequency position, the channel block of incoherent correlators is connected with n-outputs to the solver block, through which the execution results are transmitted operations of convolution of the signal sample and convolution of the noise sample in the form of an array, in the solver the received data is processed in the array, for each significant signal frequency position the rank value relative to the noise sample is calculated in accordance with the modified Mann-Whitney algorithm, the differences in the values of the signal and noise convolutions are calculated sampling, the formation of a general difference sample of the obtained convolution differences is carried out, the modulus of the value of the convolution differences is ranked, the rank value is calculated in accordance with the modified Mann-Whitney algorithm, based on the calculation of the rank of each information signal frequency position, a decision is made on the value r of binary symbols at the information output of the decision device, while the signal from the pseudo-random sequence generator is supplied to the input of the decision device and on its basis the nominal operating frequency is calculated taking into account the selected sub-band in which pseudo-random tuning of the radio frequency is performed, at each cycle of receiving frequency positions the maximum calculated rank is put in correspondence with the frequency sub-range, the obtained the maximum values of the rank of signal frequency positions and the calculated value of the nominal operating frequency represent an assessment of the quality of signal passage in the communication channel in the corresponding sub-range of operating frequencies, this assessment in the form of a two-dimensional array is input to the information and control system of the communication complex in order to provide frequency and code adaptation functions signal-code design, operational adjustment of frequency schedules compiled using radio wave propagation forecasting programs.

Images