RU2713378C1 - Method of estimating channel parameters in ofdm systems - Google Patents

Abstract

FIELD: radio equipment.SUBSTANCE: invention relates to the field of radio equipment. AFC channel estimation is carried out by means of estimation of time position and complex amplitudes of beams and is performed by means of superresolution method (MUSIC), at that initial estimation is performed by sequence of pilot symbols by means of method of least squares (LS-estimation), constructing correlation evaluation matrix, calculating eigenvectors and eigenvalues of said matrix and calculating estimates of time positions of beams, performing complex amplitude amplification of beams, channel estimate for all points is constructed, modulated data symbols are corrected.EFFECT: invention can be used in radio communication systems of multiple access with orthogonal frequency-multiplexed signals (OFDM) and is intended to increase accuracy of estimation of AFC channel.1 cl, 5 dwg

Description

The proposed method relates to the field of radio engineering and can be used in multiple access radio communication systems with orthogonal frequency-multiplexed signals (OFDM) to improve the accuracy of estimating the characteristics of the distribution channel.

In wireless communication systems, the modulated radio frequency signal, propagating from the transmitter, can reach the receiver in various ways. In OFDM systems, the bandwidth of the working range of the communication channel is divided into N orthogonal subbands, which can experience various fading effects, multipath effects, and can therefore be characterized by various complex transmission coefficients. At the same time, a complex signal is observed at the receiving side, which is the sum of copies of the transmitted signal with various unknown delays, amplitudes, and phases.

The development trend of modern radio communication systems is aimed at increasing the speed of information transfer, as well as improving the reliability of communication. Therefore, in order to meet the high requirements of modern telecommunication systems in the development of digital reception devices based on OFDM and operating in multipath conditions, a high-quality solution to the synthesis problem of the channel parameter estimation algorithm is necessary. The accuracy of such an algorithm is a determining factor in the quality of the system.

The channel condition is usually estimated by sending pilot symbols from the transmitter and measuring them at the receiver. Among the most well-known channel estimation algorithms, the maximum likelihood method (MLE) can be distinguished (see M. Morelli, U. Mengali. A Comparison of Pilot-Aided Channel Estimation Methods for OFDM Systems IEEE, Transactions on Signal Processing, vol. 49, no. December 12, 2001 pp. 3065-3073) [1] for estimating the impulse response (ИХ) of a channel in OFDM systems using pilot symbols. In accordance with the technique described in [1], the estimate of their appearance has the form

where D is a square matrix of dimension L × L

where B is a matrix of dimension N _p × L of the form

– положение n-го пилот-символа, L – длина канала (длина импульсной характеристики канала в отчетах), N – общее число поднесущих, Z – вектор значений комплексных огибащих пилот-символов. where N _p is the number of pilot symbols in one OFDM symbol,

Is the position of the nth pilot symbol, L is the channel length (the length of the channel impulse response in the reports), N is the total number of subcarriers, Z is the vector of the values of the complex envelopes of the pilot symbols.

Patent RU 2359420 H04L 25/02 [2] describes a method for estimating a radio channel in which, at the first stage, pilot symbols are corrected by regression, and at the second stage, by modulating first-order interpolation of pilot symbols, modulated data is corrected, t. to. Regression is less sensitive to the accuracy of pilot symbol estimates than interpolation.

The disadvantage of these known methods is that the resolution of the estimate of the amplitude-frequency characteristic (AFC) is not higher than the distance between the samples of the signal.

The closest analogue in technical essence to the proposed one is a channel estimation method for an OFDM communication system based on a least squares modification described in patent RU 2341023 H04B 17/00 according to claim 1 of the claims [3] adopted as a prototype.

In the prototype method, a more effective assessment of the response of the wireless communication channel is achieved due to the fact that the full frequency band of the working range of the communication channel is divided into groups of subbands. An initial channel frequency response estimate is obtained for the first group of subbands based on the transmission of the pilot symbol received in the subbands of the first group. The first group may include either all or only a subgroup of subbands used for data transmission. Then, the channel impulse response estimate is obtained based on the initial frequency response estimate and the first discrete Fourier transform matrix (DFT) for the subbands in the first group. An improved frequency response estimate for the second group of subbands is then obtained based on the impulse response estimate and the second DFT matrix for the subbands of the second group. The second group may include either all or only a subgroup of the used subbands, and should also include at least one additional subband not included in the first group if it does not include all used subbands.

An analysis of the solutions proposed in [1], [2], and [3] shows that their disadvantage is that they have a resolution of the amplitude-frequency characteristic (AFC) that is no higher than the distance between the signal samples. If the temporary position of the beam generated by the channel, that is, the reference impulse response (ИХ) of the channel, does not coincide with the temporary position of the signal reference, then the accuracy of the channel response is reduced.

In FIG. 1 and FIG. Figure 2 shows, as an example, the graphs for estimating the frequency response by the maximum likelihood method together with the true channel frequency response for a beam with a delay that coincides with the time position of the reference and does not coincide for a channel with 2 rays.

Thus, analyzing the shortcomings of the known solutions, it can be concluded that an algorithm for assessing the characteristics of the OFDM signal propagation channel, which has both high resolution and takes into account the non-stationarity of the channel, has not yet been proposed. A comparative analysis of the proposed method with the prototype, as well as other technical solutions in the art did not allow to identify the features claimed in the characterizing part of the claims.

Graphic materials used for explanation are presented in the following drawings:

FIG. 1 is a graph of the estimation of the frequency response by the maximum likelihood method together with the true frequency response of the channel for the beam with a delay coinciding with the time position of the reference for the channel with 2 rays;

FIG. 2 is a graph for estimating the frequency response by the maximum likelihood method together with the true frequency response of the channel for the beam with a delay that does not coincide with the time position of the reference for the channel with 2 rays;

FIG. 3 - a sequence of actions preceding the assessment;

FIG. 4 is a flowchart of the proposed estimation algorithm;

FIG. 5 is a simulation result of the proposed evaluation method.

The objective of the proposed technical solution is to evaluate the frequency response of the channel by evaluating the temporal position and complex amplitudes of the rays, representing the impulse response of the channel in a discrete form, and realizing it using the superresolution method (MUSIC).

(The Super Resolution Method (MUSIC) is known, see, for example, Steven M. Kay. Modern Spectral Estimation: theory & application).

To solve the problem in a method for evaluating the characteristics of a propagation channel in a receiver of a system using OFDM signals, based on sending pilot symbols from a transmitter and measuring them in a receiver, according to the invention, the frequency response of the channel is estimated by estimating the temporal position and complex amplitudes of the beams and performed using the superresolving method (MUSIC), while the initial estimation is performed according to the sequence of pilot symbols using the least squares method (LS-estimation), the correlation m evaluation matrix, calculate the eigenvectors and eigenvalues of this matrix and calculate the estimates of the temporal positions of the rays, evaluate the complex amplitudes of the rays, construct a channel estimate for all points, and correct modulated data symbols.

The main distinguishing feature of the proposed technical solution is the use of information contained in the system of eigenvectors of the correlation matrix of input signals, while the subspace of the signal sources is considered as the subspace spanned by the eigenvectors corresponding to the M largest eigenvalues, and the noise subspace as spanned by the other eigenvectors vectors.

The proposed method is carried out by a device, a diagram of which is shown in FIG. 3, where indicated:

1 - antenna;

2 - frequency converter;

3 - analog-to-digital Converter (ADC);

4 - digital signal processor (DSP).

The device contains a series-connected antenna 1, a frequency converter 2, ADC 3 and DSP 4.

The device operates as follows.

The adopted implementation of the OFDM signal is transmitted from the antenna 1 to the input of the frequency converter 2, in which the spectrum is transferred to the low-frequency region, then the signal is fed to the analog-to-digital converter 3, after which the signal samples in the form of in-phase and quadrature are received at the input of the digital signal processor 4 components. The digital processor implements the procedures of time-frequency synchronization, removal of the guard interval and fast Fourier transform (FFT), after which they proceed to the process of evaluating the channel characteristics.

The sequence of actions that immediately precede the channel characteristics estimation procedure and can be performed on the existing modern element base, for example, use the ADL5350 chip as a frequency converter, AD7767 chip as an ADC, and TMS320C6670 chip as a DSP.

. Обозначим через U длину БПФ, номера пилотных поднесущих через

,

.A data vector in the time domain is available for channel estimation. Denote it by

. We denote by U the FFT length, the numbers of pilot subcarriers by

,

.

. После этого оценка будет применена к пилотному символу (пилот-символу) по правилуThe objective of channel estimation is that, according to the received OFDM symbol X [u], it is necessary to estimate the frequency response of the channel

. After that, the score will be applied to the pilot symbol (pilot symbol) according to the rule

In FIG. 4 is a flowchart of this estimation algorithm.

Let us consider the procedure for this assessment sequentially.

In program module 5, the channel is evaluated at the pilot points using the least squares method (Hls estimate):

– заранее известные приёмной стороне значения пилотов на соответствующих поднесущих.Where

- pilot values previously known to the receiving side on the respective subcarriers.

The number of estimated rays Lest is considered known.

и окна WIND была больше, чем предполагаемое количество лучей

In program module 6, for calculating the correlation matrix, a WIND averaging window is set such that the difference in the length of the vector

and windows wind was larger than the estimated number of rays

размером

по формулеThen the matrix is built

the size

according to the formula

размером

так, что у неё стоят единицы на побочной диагонали, а все остальные элементы равны нулю. Далее строится автокорреляционная матрица по формулеand matrix

the size

so that she has units on the side diagonal, and all other elements are equal to zero. Next, an autocorrelation matrix is constructed according to the formula

,

. Собственные вектора матрицы Rxx будут записаны в матрице

по столбцам. То есть вектор W[m] будет собственным для матрицы Rxx еслиIn program module 7, the eigenvalues of the matrix Rxx are calculated and written into the real vector

,

. The eigenvectors of the matrix Rxx will be written in the matrix

by columns. That is, the vector W [m] will be eigenvalue for the matrix Rxx if

.for a fixed value

.

:After calculating the eigenvalues and eigenvectors, it is necessary to select the Lest of the largest eigenvalues. Lest of eigenvectors corresponding to Lest the largest eigenvalues in the matrix are written out

:

во втором аргументе

означает, что используются не первые Lest собственных векторов, а Lest собственных векторов, соответствующие Lest наибольшим собственным значениям.Symbol

in the second argument

means that not the first Lest eigenvectors are used, but Lest eigenvectors corresponding to Lest the largest eigenvalues.

In program module 8, three time parameters are set: the length of the observation interval

as well as the maximum and minimum values of the temporal positions of the rays Tmin and Tmax, where fs is the sampling frequency and Nfft is the FFT length.

, которое можно задавать через шаг дискретизацииThe number of sampling points of the a priori interval of possible values of the temporal positions of the rays is set

which can be set through the sampling step

.Time position estimation resolution is half the sampling step

.

размером

Table calculation

the size

Next, a valid array is calculated for each value nn = 0, ..., NN-1

лежат на отрезке

и могут выходить из него только за счёт ошибок дискретизации. Причём, в случае отсутствия шума и при бесконечно малом шаге дискретизации (то есть, при

), значения

, для которых выполняется условиеTheoretically, array values

lie on the segment

and can come out of it only due to sampling errors. Moreover, in the absence of noise and at an infinitely small sampling step (i.e., at

), values

for which the condition

. Условие яркой выраженности максимумов необходимо учитывать, используя субъективный подход. correspond to the temporary positions of the rays. In the case of noise, the estimates of the temporal positions of the rays will correspond to the positions of pronounced array maxima

. The condition for the pronounced severity of the maxima must be taken into account using a subjective approach.

. Значения оценок временных положений лучей в секундахDenote the Lest indices of the largest pronounced maxima by

. The values of the estimates of the temporal positions of the rays in seconds

Tasks for estimating the temporal positions of rays in samples

This completes the assessment of the temporal positions of the rays.

,

и LS-оценка канала, построенная по пилот-символам. Оценка амплитуд лучей производится в соответствии с методом максимального правдоподобия. Комплексную амплитуду l-го луча обозначим

At the stage of evaluating the complex amplitudes of the rays (program module 9), an estimate of the number of rays Lest and estimates of the temporal positions of the rays are known

,

and LS channel estimation based on pilot symbols. The amplitudes of the rays are estimated in accordance with the maximum likelihood method. We denote the complex amplitude of the lth ray

и

. Одномерная оценка ЧХ канала строится по формулеAt the stage of constructing the channel estimate for all points (block 10), the Lest estimates are known,

and

. One-dimensional estimation of the frequency channel is constructed according to the formula

– номера поднесущих, используемых для передачи всего сигнала.Where

- subcarrier numbers used to transmit the entire signal.

In program module 11, channel estimation is calculated according to the rule

The effectiveness of the proposed solution was tested in the modeling process. During the simulation, the following parameters were selected:

- size of the OFDM symbol (DFT duration) - 2048;

- the number of pilot symbols is 128;

- the true number of channel rays is 2 or 3;

- the temporary position of the beam does not necessarily coincide with the temporary position of the signal reference;

- sampling frequency - 48 kHz;

- WIND averaging window - 32;

- the number of sampling points of the a priori interval of possible values of the temporal positions of the rays is 10000.

In FIG. Figure 5 shows the frequency response at the pilot points of the true channel and channel after estimation by the superresolution method. As can be seen from FIG. 5, the proposed algorithm makes it possible to estimate the frequency response of the channel by estimating the temporal position of the rays for the case where the temporal position of the beam does not coincide with the temporal position of the signal reference.

Thus, in the proposed method, in contrast to the already known methods, the frequency response of the channel is estimated not only by the pilot symbols at the points corresponding to the signal samples, but also with high resolution, the temporal position and amplitudes of the rays are estimated using all a priori knowledge as well as information contained in the system of eigenvectors of the correlation matrix of input signals, with maximum efficiency.

Claims (1)

A method for evaluating the characteristics of a propagation channel in a receiver of a system using OFDM signals based on sending pilot symbols from a transmitter and measuring them in a receiver, characterized in that the channel frequency response is estimated by estimating the temporal position and complex amplitudes of the rays and is performed using the superresolving method ( MUSIC), in this case, the initial assessment is performed according to the sequence of pilot symbols using the least squares method (LS-estimate), the correlation matrix of the estimate is built, and the These vectors and eigenvalues of this matrix also calculate the estimates of the temporal positions of the rays, estimate the complex amplitudes of the rays, construct a channel estimate for all points, and correct modulated data symbols.

Images