RU2398359C2 - Method for transmission-reception of signal in multiuser system of radio communication with multiple transmitting and multiple receiving antennas - Google Patents

Abstract

FIELD: radio engineering.

SUBSTANCE: invention applies new sequence of interrelated actions, including procedure of vector disturbance in combination of array basis reduction and multi-alternative quantisation. Invention makes it possible to simultaneously service group of several subscriber stations in one and the same physical channel. Invention advantage is possibility of quite simple realisation in transmitter and especially simple realisation in receiver of subscriber station. Invention advantage is possibility of realisation with only one receiving antenna available in each of subscriber stations.

EFFECT: increased throughput capacity of communication channel.

6 cl, 7 dwg

Description

The invention relates to the field of radio engineering, in particular to a method for transmitting and receiving a signal in a multi-user radio communication system with multiple transmitting and multiple receiving antennas.

The technology of using several transmitting and several receiving antennas attracts attention as an effective way to increase the bandwidth of the communication channel, which does not require additional costs of the radio frequency spectrum. In radio communication systems using this technology, the communication channel between the transmitting and receiving sides has multiple inputs — transmitting antennas — and multiple outputs — receiving antennas, as a result of which the technology is called MIMO (multiple-input-multiple -output).

The whole set of signal propagation channels between transmitting and receiving antennas is called a MIMO channel. One way to increase throughput is to simultaneously transmit various information streams over different spatial subchannels of the MIMO channel. This method is known as spatial multiplexing [1] GJFoshini, GDGolden, RAValenzuela, "Simplified processing for high spectral efficiency wireless communication employing multi-element arrays," IEEE Selected Areas Communication, vol.17, pp. 1841 -1852, November, 1999, [2] 802.16TM IEEE Standard for local and metropolitan area networks. Part 16: Air Interface for Fixed Broadband Wireless Access Systems, October 1, 2004.

In spatial multiplexing, independent information streams are transmitted through various transmit antennas. On the receiving side, the transmission coefficients h _{j, i of} all spatial communication channels are estimated, each of which is formed by one transmitting and one receiving antenna, where i, j are the indices of the transmitting and receiving antennas, respectively. Of these coefficients form a channel matrix H, which is used when receiving a signal.

Until recently, transmission-reception methods for single-user MIMO channels, covering one receiver and one transmitter (in terms of foreign publications, point-to-point, from point to point), have been very intensively developed.

One of the most serious obstacles to using MIMO technology in a point-to-point system is the need to place several antennas at a subscriber station (AC). This is quite difficult to implement, since the requirements for small dimensions and low cost are usually imposed on the subscriber station.

Another problem with using single-user MIMO technology is that the increase in throughput depends on the scattering properties of the signal propagation medium. Moreover, to obtain a significant gain in throughput, it is necessary that the signal propagation medium has scattering objects, and the antenna systems have antennas that are distant from each other over a large distance.

An option to solve these problems is the multi-user MIMO technology. In this technology, the channel formed by several antennas of a base station (BS) on the one hand and antennas of several subscriber stations (AS) on the other hand is considered as a MIMO channel. At the same time, each subscriber station can have both several and only one antenna.

Multi-user approaches provide the opportunity to take advantage of the additional benefits of MIMO technology.

Firstly, it becomes possible to increase throughput due to the spatial separation of users when several subscriber stations use the same physical channel to communicate with the BS.

Secondly, the multi-user MIMO channel has a relatively low correlation between spatial subchannels, due to the fact that they belong to different user terminals. This provides bandwidth gains even in low dispersion environments.

Thirdly, it becomes possible to implement MIMO algorithms when the subscriber equipment has one or a small number of antennas.

To date, there is a fairly practical solution for the multi-user MIMO algorithm in the return channel of the communication system (from subscriber stations to the base). This is a collaborative spatial multiplexing method used to transmit signals from several subscriber terminals to the base station. This solution is provided by modern communication standards, for example [2] 802.16ТМ IEEE Standard for local and metropolitan area networks. Part 16: Air Interface for Fixed Broadband Wireless Access Systems, October 1, 2004.

However, the problem of increasing capacity is most relevant for the direct channel - from the base station to the subscriber terminals, through which the most voluminous and high-speed data streams are transmitted. At the same time, a simple and effective multi-user algorithm for the direct channel of the MIMO communication system has not yet been developed. The implementation of multi-user MIMO approaches in the direct channel faces two main problems. First of all, it is the need to provide the transmitter with information about the communication channel. Another problem is that, unlike a single-user MIMO channel, in a multi-user channel it is practically impossible to jointly process signals from different user terminals.

Thus, the task of developing a multi-user signal transmission-reception algorithm in the direct channel of the MIMO communication system is very urgent.

There are several multi-user MIMO approaches in the forward channel. These include “dirty paper coding” [3] M.Airy, A. Forenza, RWHeath, Jr.S.Shakkottai, "Practical Costa preceding for the multiple antenna broadcast channel," IEEE Global Telecommunications Conference, GLOBECOM, 29 Nov.-3 Dec. 2004, Volume 6, Page (s): 3942-3946, block diagonalization [4] QHSpencer, and M.Haardt, "Capacity and Downlink Transmission Algorithms for a Multi-user MIMO Channel," Signals, Systems and Computers, 2002. Conference Record of the Thirty-Sixth Asilomar Conference, Volume 2, Issue, 3-6 Nov. 2002 Page (s): 1384-1388 vol.2, and various linear multiuser precoding methods [5] JCMundarath, JHKotecha, "Zero-Forcing Beamfbrming for Non-Collaborative Space Division Multiple Access," Proceedings of 2006 IEEE International Conference on Acoustics, Speech and Signal Processing ICASSP, 14-19 May 2006, Volume: 4, page (s): IV-IV. [6] A Wiesel, Y. C. Eldar, and Sh. Shamai, "Optimal Generalized Inverses for Zero Forcing Preceding," 41st Annual Conference on Information Sciences and Systems, CISS'07, March 14-16, 2007, pages 130-134.

Most of these methods have a high implementation complexity and require serious research aimed at their practical application.

Known, for example, is the block diagonalization algorithm, which theoretically is a very effective way to implement the MIMO multi-user technology [4] Q. H. Spencer, and M. Haardt, "Capacity and Downlink Transmission Algorithms for a Multi-user MIMO Channel, "Signals, Systems and Computers, 2002. Conference Record of the Thirty-Sixth Asilomar Conference, Volume 2, Issue, 3-6 Nov. 2002 Page (s): 1384-1388 vol. 2. In this algorithm, multi-user pre-conversion (coding) of the signal is performed in such a way that the MIMO channel is transformed into orthogonal spatial subchannels corresponding to different user terminals. However, these channels do not cause mutual interference. Reception and transmission of signals for each subscriber terminal is performed in the corresponding spatial subchannel using any of the known single-user MIMO algorithms.

To implement this approach, it is necessary to evaluate the transmission coefficients of all spatial communication channels and form a channel matrix. Information about the channel matrix is an auxiliary control information, which in one way or another must be transmitted to the base station. After that, the base station must decompose the channel matrix into singular values. The resulting information about the right singular vectors BS uses in the process of signal transmission. In this case, the base station must transmit information about the left singular vectors to the subscriber terminals so that they can receive the signal.

Such an algorithm is difficult for practical implementation, since it requires two-way transmission of very voluminous control data at high speed. Another disadvantage of this algorithm is that it is applicable only for the case when subscriber terminals have two or more receiving antennas.

Simpler linear methods of multi-user precoding are known, which include the minimum mean squared error (MMSE) method and the zero forcing (ZF) method [5] JCMundarath, JHKotecha, "Zero-Forcing Beamforming for Non-Collaborative Space Division Multiple Access, "Proceedings of 2006 IEEE International Conference on Acoustics, Speech and Signal Processing ICASSP, 14-19 May 2006, Volume: 4, page (s): IV-IV, [6] A Wiesel, YCEldar, and Sh.Shamai, "Optimal Generalized Inverses for Zero Forcing Preceding," 41st Annual Conference on Information Sciences and Systems, CISS'07, March 14-16, 2007, pages: 130-134.

In these algorithms, the signal processing (precoding) preceding the transmission is performed by linear transformation, the matrix of which is formed on the basis of the inverse or pseudo-inversion of the channel matrix H. As a result of such preliminary processing, only a useful signal intended for this antenna is generated without interference from each antenna, generated by signals from other receiving antennas. The ZF and MMSE methods are applicable to terminals equipped with one or several antennas.

One of the simplest methods of multi-user precoding is the channel inversion or zeroing (ZF) method.

According to the channel inversion method, from the modulation symbols α ₁ , ..., α _M intended for simultaneous transmission to the subscriber terminals, a packet or vector of modulation symbols a = [α ₁ , ... α _M ] ^T is formed in which the number of symbols transmitted by each speaker, equal to the number of receiving antennas of this AS, where M is the total number of receiving antennas of subscriber stations. A vector of transmitted signals s is formed from this vector by multiplying the vector a by the inverse of the channel matrix or pseudo-inversion if the matrix H is not square. Below, for simplicity, we consider the case M = N, when the matrix H is square. Then

Many signals received by subscriber stations can be represented as elements of the vector y, which, in turn, can be expressed as

where n is the vector of noise components of the receiving antennas, which are well approximated as independent Gaussian random variables, x is the normalized vector of transmitted signals obtained by the following transformation of the vector s:

- мощность сигнала, E[γ] - матожидание γ.

is the signal power, E [γ] is the expectation of γ.

Substituting (1) and (3) into (2), we can obtain

where n is the vector of noise components of the receivers AC, I _M is the unit diagonal matrix of dimension M × M.

From the formula (4) it is seen that the received user signals are mutually independent and do not create mutual interference. However, normalization (3) leads to the fact that the signal transmission coefficient is

в знаменателе этого выражения зависит от инверсии канальной матрицы Н и может быть весьма значительной, особенно при плохо обусловленной канальной матрице. Наличие этого коэффициента является основной причиной снижения относительной полезной мощности в точке приема и, вместе с этим, помехоустойчивости приема.Value

in the denominator of this expression depends on the inversion of the channel matrix H and can be very significant, especially with a poorly conditioned channel matrix. The presence of this coefficient is the main reason for the decrease in the relative net power at the receiving point and, along with this, the noise immunity of the reception.

Thus, a significant increase in signal power s due to multi-user pre-processing is the main disadvantage of ZF and MMSE methods. Since there is a limitation on the transmission power in the communication system, the signal amplitude is linearly reduced (in accordance with (3)), however, this leads to a significant decrease in the useful signal power relative to the noise at the receiving point. As a result, reception noise immunity becomes low.

There is another way to limit the transmit power, which avoids a significant reduction in the relative net power at the receiving point. The method is based on the operation of nonlinear modular reduction, which is used, for example, in [7] R.F.H. Fischer, C. Windpassinger, A. Lampe, J.B. Huber, "Space-Time Transmission using Tomlinson-Harashima Preceding," In Proc. 4th Int. ITG Conf., Pp. 139-147, Berlin, Jan. 2002.

The input quantity for this operation is a complex number that reflects the converted signal. The operation of modular reduction (modulo reducing) consists in adding to the real and imaginary parts of the input number of quantities that are multiples of the real value A, called the module.

The added values are selected so that the total complex number appears in the central region of the complex plane, in which all the complex symbols of the modulation constellation used are located. The magnitude of the module is known to both the transmitting and receiving sides, which allows you to restore the reduced signal during reception.

The most efficient way to use nonlinear modular reduction is the vector perturbation algorithm [8] Christoph Windpassinger, Robert FH Fischer, and Johannes B. Huber, "Lattice-Reduction-Aided Broadcast Precoding," IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 52 , NO.12, DECEMBER 2004, pp.2057-2060.

The vector perturbation consists in adding to the vector of information symbols a some perturbing vector p. As a result, the signal after multi-user conversion can be represented as

The real and imaginary parts of the elements of the vector p are determined in multiples of the value of the module A, chosen in such a way that

where Reα, Imα are the real and accordingly imaginary parts of any complex symbol of the modulation constellation used.

The signal γ received in the channel of each receiving antenna of each subscriber station is subjected to non-linear modular reduction operations

Where

[x] is the maximum integer not exceeding x.

The main property of this operation is that it is invariant to the addition of quantities that are multiples of A:

where r is any integer.

Due to this property, after modular reduction, the signals of the receiving antennas of all speakers can be represented by a vector

where I _M is the unit diagonal matrix of dimension M × M.

This equality shows that the vectors of the transmitted and received signals are connected linearly using the diagonal matrix I _M. That is, as a result of the previous transmission of the multi-user transform, a corresponding transmitted signal is generated in each of the receiving antennas without interference caused by signals transmitted to other receiving antennas.

Equality (10) is obtained under the assumption that the noise-distorted symbols of the modulation constellation do not go beyond the square of the complex plane, limited by the values

i.e:

where Ren, Imn is the real and, accordingly, imaginary parts of the noise component of the signal of the receiving antenna.

In cases where condition (11) is not fulfilled, nonlinear modular reduction (10) causes signal distortion, which, in turn, leads to a loss of noise immunity and, accordingly, communication channel capacity. Therefore, it is desirable to minimize the transmitted signal power x = H ^-1 · (a + p). For this, it is necessary to determine the optimal disturbance vector p _opt in such a way that adding it to the vector of information symbols a will provide a minimum signal power after preliminary multi-user encoding:

- множество М-мерных векторов, элементы которых имеют целочисленные действительную и мнимую часть.Where

- a set of M-dimensional vectors whose elements have integer real and imaginary parts.

бесконечно. Поэтому поиск решения методом перебора всех значений множества

невозможен. Даже если ограничить множество рассматриваемых целых чисел несколькими наиболее близкими к нулю значениями, например {-2, -1, 0, 1, 2}, то и в этом случае множество поиска может быть весьма велико. Например, это множество состоит из (5²)^M=625 векторов при М=2, и (5²)^M=390625 векторов при М=4.The solution to the optimization problem (12) is complicated by the fact that the set of integers is unbounded, and therefore the set

infinitely. Therefore, finding a solution by enumerating all the values of the set

impossible. Even if we limit the set of integers under consideration to several values closest to zero, for example {-2, -1, 0, 1, 2}, then in this case the search set can be very large. For example, this set consists of (5 ² ) ^M = 625 vectors at M = 2, and (5 ² ) ^M = 390625 vectors at M = 4.

Therefore, the enumeration method for solving (12) leads to a significant increase in the complexity of implementation.

One approach to solving optimization problem (12) is to use lattice basis reduction [8] Christoph Windpassinger, Robert F.H. Fischer, and Johannes B. Huber, "Lattice-Reduction-Aided Broadcast Preceding," IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 52, NO.12, DECEMBER 2004, pp.2057-2060. This method is the closest to the method of the claimed invention. The prototype method is as follows.

A method of transmitting and receiving a signal in a radio communication system including a transmitting station equipped with N transmitting antennas and K receiving stations, where K≥2, with each receiving station equipped with at least one receiving antenna, and the total number of receiving antennas of receiving stations M satisfies the condition 1 <M≤N, namely, that

evaluate the parameters of the set of spatial communication channels, each of which is formed by one transmitting antenna of the transmitting station and one receiving antenna of the receiving station,

as the estimated parameters using the transmission coefficients of the communication channel;

transmit-receive signals between the transmitting station and the receiving stations, for which:

- At the transmitting station, K sets of modulation symbols are formed for transmission to the receiving stations, respectively,

- from K formed sets of modulation symbols form packets of M modulation symbols in each, one packet symbol for each of the receiving antennas of the receiving stations,

- the modulation symbol packet is represented as a vector of transmitted modulation symbols a = [α ₁ ... α _M ] ^T ,

- perform multi-user conversion of the vector of transmitted modulation symbols a into a vector of transmitted signals x so that the transmitted signals do not cause mutual interference in the M receiving antennas of the receiving stations, for which

form the channel matrix H using the transmission coefficients of the spatial communication channels,

from the vector of transmitted modulation symbols a and the channel matrix H, real-valued vector a _r and matrix H _{r are formed} in accordance with the formulas

where ReY, ImY are matrices composed of real and respectively imaginary parts of the corresponding elements of the matrix Y,

from a real-valued channel matrix H _r form a matrix W _r preliminary signal conversion,

by reducing the lattice basis of the matrix W _r , an integer matrix T with a determinant of ± 1 is formed, multiplication by which converts the preliminary transformation matrix into the matrix Z = W _r T, which has a knowingly low condition number,

using the matrix T, the perturbing vector is determined by the formula

where Q (x) is the vector obtained from the vector x by rounding its elements to the nearest integers,

A is a real number such that the real Reα and imaginary Imα parts of any modulation symbol in absolute value are strictly less than A / 2:

form a perturbed real-valued vector of the transmitted modulation symbols by summing the real-valued vector of the transmitted modulation symbols and the perturbing vector a _r + p ₀ , and perform preliminary linear conversion of the resulting perturbed real-valued vector of the modulation symbols, thus forming a real-valued vector transmitted signals:

from the obtained real-valued vector of transmitted signals x _r form a vector of transmitted signals

where j is the imaginary unit, and x _r (n: m) denotes a vector composed of a sequence of elements of the vector x _r from the n-th to the m-th;

- the set of signals corresponding to the elements of the vector of transmitted signals x is transmitted through all transmitting antennas - one signal through the antenna;

- receive signals at each of the K receiving stations, and the reception is carried out in the channel of each receiving antenna, and in the process of reception

generating a signal y as a complex number with a module and an argument reflecting, respectively, the amplitude and phase of the signal received by the channel of a given antenna;

determine the real and imaginary parts of the normalized signal y

with the received signals z and c perform the operation of modular reduction with a module equal to A:

where [x] is the integer part of x, that is, the maximum integer not exceeding x,

и

формируют комплексный сигналfrom signals

and

form a complex signal

сформированные таким образом в каждом физическом канале каждой приемной антенны, выполняют демодуляцию и декодирование принятого сигнала.using complex signal values

thus formed in each physical channel of each receiving antenna, demodulate and decode the received signal.

This method of transmitting and receiving a signal in a multi-user communication system MIMO uses preliminary linear signal conversion based on the inverse (or pseudo-inversion) of the channel matrix.

This is a very effective method of multi-user precoding, since, firstly, as a result of such a linear conversion, the mutual interference of signals in the receiving antennas is suppressed. Secondly, the receiving side does not require any additional overhead information to demodulate the signal, as a result of which a relatively simple implementation of the receiving device is possible.

However, by multiplying the signal by the inversion (or pseudo-inversion) of the channel matrix, the signal power is significantly increased. Both in the claimed method and in the prototype method, a vector disturbance procedure is used to reduce power. This procedure consists in adding a certain disturbing vector to the signal.

The elements of the perturbing vector are multiples of the quantity A, which is known to both the transmitter and the receiver. The value of A is determined depending on the type of modulation used, so that it is possible to restore the original signal in the receiver using the operation of modular reduction.

Чем меньше величина

тем меньше мощность передачи и, вместе с тем, меньше степень искажения сигнала за счет модульного редуцирования в приемнике.The optimal perturbing vector is selected from the set of vectors p = A · Z ^2M , where Z ^{2M is a} discrete set of integer vectors of dimension 2M. Moreover, the optimal disturbing vector is defined as a vector that minimizes the value

Lower value

the lower the transmit power and, at the same time, the less the degree of signal distortion due to modular reduction in the receiver.

The simplest way to determine the optimal perturbation vector is to first determine the non-quantized perturbation vector p _u as close as possible to a _r

and then quantize it

where Q (x) is the element-wise rounding to the nearest integer.

However, in the presence of even small noise distortions in the vector a _{r, the} rounding operation leads to even greater distortions of the vector W _r · p. This phenomenon is called noise amplification due to quantization. This noise gain depends on the degree of orthogonality of the columns of the matrix W _r . The higher the degree of orthogonality of the columns of the matrix W _r , the less noise distortion of the vector W _r · p.

Therefore, to reduce distortion using the method of reduction of the basis of the lattice. In this case, the preliminary linear transformation matrix W _{r is} transformed into the matrix Z, which has a knowingly low condition number and, therefore, a higher degree of orthogonality of the columns. Such a transformation is performed by reducing the lattice basis, that is, in such a way that the relation Z = W _r T holds between the original and the transformed matrix, where T is an integer matrix with a determinant of ± 1.

After that, the non-quantized perturbation vector is determined in the space of the reduced matrix Z, i.e.

Where

The optimal disturbing vector is found by quantization

and subsequent transformation, inverse reduction of the lattice basis

чем вектор, полученный без редукции базиса решетки (22).This perturbing vector provides a smaller value

than the vector obtained without reduction of the lattice basis (22).

Это обусловливает, во-первых, увеличение диапазона значений передаваемой мощности сигнала и, во-вторых, увеличение искажений сигнала в приемнике в процессе нелинейного модульного редуцирования. Первый из этих аспектов приводит к тому что, увеличивается отношение пиковой мощности сигнала к средней, что повышает требования к линейности усилителя и затрудняет реализацию способа в аппаратуре связи. Второй аспект вызывает снижение пропускной способности канала.However, despite the fact that the transformation of the reduction of the lattice basis on average reduces the condition number of the matrix and increases the degree of orthogonality of its columns, it does not guarantee the ideal orthogonality of the columns of the matrix of the preliminary linear transformation. As a consequence, the perturbing vector chosen in this way does not always provide a minimum value

This causes, firstly, an increase in the range of the transmitted signal power and, secondly, an increase in signal distortion in the receiver during non-linear modular reduction. The first of these aspects leads to the fact that the ratio of peak signal power to average increases, which increases the requirements for the linearity of the amplifier and complicates the implementation of the method in communication equipment. The second aspect causes a decrease in channel capacity.

The problem that the claimed invention solves is to increase the throughput of the communication channel, which is achieved by the claimed method by using a new sequence of interrelated actions, including the vector perturbation procedure in combination with the reduction of the lattice basis and multi-alternative quantization.

A method of transmitting and receiving a signal in a multi-user radio communication system with multiple transmitting and multiple receiving antennas, wherein a transmitting station equipped with N transmitting antennas and K receiving stations are used, where K≥2, wherein each receiving station is equipped with at least one receiving antenna, and the total number of receiving antennas of the receiving stations M satisfies the condition 1 <M≤N, while the transmission-reception of signals between the transmitting and receiving stations is carried out through F physical communication channels, de F≥1, consists in the fact that

for each of the F physical channels, the parameters of the set of spatial communication channels are evaluated, each of which is formed by one transmitting antenna of the transmitting station and one receiving antenna of the receiving station;

transmit-receive signals between the transmitting station and the receiving stations using F physical channels, for which:

- At the transmitting station, K sets of modulation symbols are formed for transmission to the receiving stations, respectively,

- from K generated sets of modulation symbols, F packets of M modulation symbols in each are formed, including in a packet of M _k modulation symbols for each k-th receiving station, where M _k is the number of receiving antennas of the k-th receiving station,

- carry out the transmission of each of the F packets of modulation symbols on the corresponding physical channel, while

the modulation symbol packet is represented as a vector of transmitted modulation symbols a = [α ₁ , ..., α _M ] ^T , each element of which is a complex number, with a modulus and argument reflecting the amplitude, and accordingly the phase of the corresponding modulation symbol,

carry out multi-user conversion of the vector of transmitted modulation symbols a into a vector of transmitted signals x so that the transmitted signals do not cause mutual interference in the M receiving antennas of the receiving stations, for which

form a channel matrix H for a given physical channel using the transmission coefficients of the spatial communication channels,

from the vector of transmitted modulation symbols a and the channel matrix H, real-valued vector a _r and matrix H _{r are formed} in accordance with the formulas

where ReY, ImY are matrices composed of real and respectively imaginary parts of the corresponding elements of the matrix Y,

from a real-valued channel matrix H _r form a matrix W _r preliminary linear signal conversion,

by reducing the lattice basis of the matrix W, an integer matrix T is formed with a determinant of ± 1, multiplication by which converts the preliminary linear transformation matrix into a matrix Z = W _r T having a knowingly low condition number,

using the matrix T and the real-valued vector of the transmitted modulation symbols a _r , determine the non-quantized perturbation vector as

where A is a real number, and such that the real and imaginary part of any modulation symbol in absolute value is less than A / 2,

rounding off each of the elements of the obtained vector z to the nearest integer in magnitude, thus determining the first quantized vector z ₁ , and determining the vector of the corresponding quantization error values:

where Q (z) is the vector obtained by elementwise rounding of the vector z to the nearest integer,

form the second quantized vector z ₂ by determining for each element of the vector z the second closest integer with the opposite quantization error value, as well as a second vector of the corresponding quantization error values,

from the elements of the first z ₁ and second z ₂ quantized vectors form R quantized vectors u having the smallest values of the total quantization error of the vector,

each of R quantized vectors u is transformed by the formula

thus forming many candidate disturbing vectors,

минимальна,determine the optimal disturbing vector p ₀ as a vector from the set of candidate disturbing vectors for which the decisive function

minimal

form a perturbed real-valued vector of the transmitted modulation symbols by summing the real-valued vector of the transmitted modulation symbols and the optimal perturbing vector, and perform preliminary linear transformation of the resulting perturbed real-valued vector of the modulation symbols, thus forming a real-valued vector of the transmitted signals:

from the received real-valued vector of the transmitted signals x _r form an unnormalized vector of transmitted signals

where j is the imaginary unit,

and x _r (n: m) denotes a vector composed of a sequence of elements of the vector x _r from n-th to m-th;

form a vector of transmitted signals x, multiplying the vector of unnormalized transmitted signals by a normalization coefficient C _T ,

the set of signals corresponding to the elements of the obtained vector x is transmitted in the corresponding physical channel through all the transmitting antennas - one signal through the antenna;

- receive signals at each of the K receiving stations, and in each physical channel of each receiving antenna, the reception is carried out in such a way that

generating a signal y as a complex number with a module and an argument corresponding to the amplitude and phase of a signal received by a given physical channel;

normalize the signal y, multiplying it by the normalization coefficient C _R , thus forming a normalized signal

the coefficient of regulation C _R set, for example, equal to the reciprocal of the coefficient of regulation of transmission:

determine the real and imaginary parts of the normalized signal y _norm

with the received signals z and c perform the operation of modular reduction modulo A:

where [x] is the integer part of x, that is, the maximum integer less than x,

и

формируют комплексный сигналfrom signals

and

form a complex signal

, сформированные таким образом в каждом физическом канале каждой приемной антенны, выполняют демодуляцию и декодирование принятого сигнала.using complex signal values

thus formed in each physical channel of each receiving antenna, demodulate and decode the received signal.

Moreover, for example, for each of the F physical channels, the transmission coefficient of the communication channel and the signal-to-noise ratio in the channel are used as estimated parameters.

At the transmitting station, each of the K information messages intended for transmission to the receiving stations, for example, is represented respectively as a sequence of binary symbols, and then the coding, interleaving and modulation of the binary symbols of this sequence are performed.

The matrix W _r preliminary linear signal conversion form, for example, as

where H _r is a real-valued channel matrix for a given physical channel.

The second quantized vector z ₂ and the second vector of the corresponding quantization error values are formed by the formulas:

where sign (a) denotes the vector obtained from the vector a by applying the operation to each of its elements:

The total quantization error of a vector is determined, for example, as the sum or sum of squares of the absolute quantization errors of all elements of the vector.

When forming the vector of the transmitted signals, the normalization coefficient C _{T is} chosen so that the average transmit power of the signals of the generated vector x is equal to the power of the signals transmitted to the receiving stations without multi-user conversion. Moreover, these may be, for example, pilot signals used for channel estimation at the receiving side.

The inventive method of transmitting and receiving a signal in a multi-user radio communication system with multiple transmitting and multiple receiving antennas in comparison with the prior art has novelty. Distinctive features of the invention are the following features:

transmission-reception of signals between transmitting and receiving stations is carried out by means of F physical communication channels, where F≥1,

for each of the F physical channels, the parameters of the set of spatial communication channels are estimated,

transmitting and receiving signals between the transmitting station and the receiving stations using F physical channels,

from K formed sets of modulation symbols form F packets of M modulation symbols in each,

transmit each of the F modulation symbol packets on a respective physical channel,

defining a non-quantized perturbation vector. use a real-valued vector of transmitted modulation symbols a _r , while

the non-quantized perturbation vector is defined as

where A is a real number, and such that the real and imaginary part of any modulation symbol in absolute value is less than A / 2,

rounding off each of the elements of the obtained vector z to the nearest integer in magnitude, thus determining the first quantized vector z ₁ , and determining the vector of the corresponding quantization error values:

where Q (z) is the vector obtained by elementwise rounding of the vector z to the nearest integer,

form the second quantized vector z ₂ by determining for each element of the vector z the second closest integer with the opposite quantization error value, as well as a second vector of the corresponding quantization error values,

from the elements of the first z ₁ and second z ₂ quantized vectors form R quantized vectors u having the smallest values of the total quantization error of the vector,

the total quantization error of the vector is determined, for example, as the sum or sum of squares of the absolute quantization errors of all elements of the vector,

each of R quantized vectors u is transformed by the formula

thus forming many candidate disturbing vectors,

минимальна,determine the optimal disturbing vector p ₀ as a vector from the set of candidate disturbing vectors for which the decisive function

minimal

form a perturbed real-valued vector of the transmitted modulation symbols by summing the real-valued vector of the transmitted modulation symbols and the optimal perturbing vector, and perform preliminary linear transformation of the resulting perturbed real-valued vector of the modulation symbols, thus forming a real-valued vector of the transmitted signals,

form a vector of transmitted signals, multiplying the vector of unnormalized transmitted signals by a normalization coefficient C _T ,

receiving signals at each of the K receiving stations, in each physical channel of each receiving antenna, the reception is carried out in such a way that the generated signal is normalized, multiplying it by the normalization coefficient C _R , thus forming a normalized signal

the coefficient of regulation C _R set, for example, equal to the reciprocal of the coefficient of regulation of transmission:

The essential distinguishing features listed above allow you to get the best technical effect, namely:

- significantly increase the throughput of the multi-user MIMO communication system, since a simultaneous service of a group of several subscriber stations in the same physical channel is performed;

- surpass the well-known multi-user MIMO algorithms [4] QHSpencer, and M.Haardt, "Capacity and Downlink Transmission Algorithms for a Multi-user MIMO Channel," Signals, Systems and Computers, 2002. Conference Record of the Thirty-Sixth Asilomar Conference, Volume 2, Issue, 3-6 Nov. 2002 Page (s): 1384-1388 vol.2, [5] JC Mundarath, JHKotecha, "Zero-Forcing Beamforming for Non-Collaborative Space Division Multiple Access," Proceedings of 2006 IEEE International Conference on Acoustics, Speech and Signal Processing ICASSP, May 14-19, 2006, Volume: 4, page (s): IV-IV, [6] A Wiesel, YCEldar, and Sh. Shamai, "Optimal Generalized Inverses for Zero Forcing Preceding," 41st Annual Conference on Information Sciences and Systems, CISS '07, March 14-16, pages 130-134, as well as the algorithm that prototyped the claimed method [8] Christoph Windpassinger, Robert FH Fischer, and Johannes B. Huber, "Lattice-Reduction-Aided Broadcast Preceding, "IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 52, NO.12, DECEMBER 2004, pp.2057-2060.

This advantage of the proposed method is achieved by using a new sequence of interrelated actions, including the vector perturbation procedure in combination with the reduction of the lattice basis and multi-alternative quantization.

In addition, the advantage of the method according to the claimed invention is the possibility of its relatively simple implementation in the transmitter and, especially, simple implementation in the receiver of the subscriber station.

A significant advantage of this invention is that it is possible to implement if there is only one receiving antenna at each of the subscriber stations of the communication system.

Another important advantage of the claimed invention is the possibility of its implementation in almost any distribution environment. Let us explain that for the implementation of traditional single-user MIMO methods, a propagation medium with a large number of scattering objects is necessary, which is not always implemented in practice. At the same time, the method according to the claimed invention provides a gain in throughput even in a relatively low scattering environment, since the receiving side antennas belong to different user terminals, as a result of which their signals have a low correlation regardless of the properties of the signal propagation medium.

Further, the description of the invention is illustrated by examples and drawings.

Figure 1 is a structural diagram of a multi-user radio communication system with multiple transmitting and multiple receiving antennas, in which the inventive method is carried out.

Figure 2 is a structural diagram of the transmitter of a multi-user radio communication system MIMO-OFDM.

Figure 3 is a structural diagram of a signal shaper of a group of AC joint service used in the transmitter.

Figure 4 - structural diagram of the node forming the information packets used in the shaper of the signal group AC joint service.

Figure 5 is a structural diagram of a multi-user processing node used in the signal shaper of the joint service group AS.

Figure 6 is a structural diagram of a speaker receiver.

7 is a structural diagram of a signal processing unit of a receiving antenna used in the receiver of the speaker.

On Fig - noise immunity characteristics of multi-user MIMO algorithms.

Structural diagrams of devices and their constituent blocks, made in figure 1-7, are shown as examples for the implementation of the proposed method. However, the use of the claimed invention is not limited to its implementation only through the above devices.

The method of transmitting and receiving signals in a multi-user radio communication system with multiple transmitting and multiple receiving antennas according to the claimed invention is implemented in a system that contains a base station (BS) and at least two subscriber stations (AS). The structure of such a radio communication system consisting of one BS 1 and L of subscriber stations 6.1-6.L is shown in FIG. The base station 1 equipment includes at least a transmitter 2, a receiver 3, and a control unit 4, the first input and output of the control unit 4 being connected respectively to the output and input of the transmitter 2, the second input and output of the control unit 4 being connected respectively to the output and the input of the receiver 3. The equipment of each subscriber station 6.1-6.L also includes at least a receiver, respectively 8.1-8.L, a transmitter 9.1-9.L and a control unit 10.1-10.L, the first input and the output of the control units 10.1-10.L are connected respectively to the output m and the input of the receivers 8.1 - 8.L, the second input and output of the control units 10.1-10.L are connected respectively to the output and input of the transmitters 9.1-9.L.

The communication channel from transmitter 2 BS 1 to receivers 8.1-8.L AC 6.1-6.L is usually called the direct channel of the communication system, and the communication channel from transmitters 8.1-8.L AC 6.1-6.L to receiver 3 .

BS 1 is equipped with N antennas 5.1-5.N, the inputs and outputs of which are connected to the inputs and outputs of the transmitter 2 and receiver 3, and which are used to transmit the signal in the forward channel. The same N antennas 5.1-5.N can also be used to receive a reverse channel signal, with N> 1.

Each i-th AC (i = 1, ... L) from L AC 6.1-6.L is equipped with 7.1-7.M _i transceiver antennas, the outputs and inputs of which are connected to the inputs and outputs of the corresponding transmitters 9.1-9.L and receiver 8.1-8.L. The number of antennas M _i for different speakers (i = 1, ... L) can be different. In this case, the system can include both speakers equipped with one antenna and speakers equipped with several antennas, that is, N≥M _i ≥1. In the particular case, all subscriber stations can have one antenna.

The claimed invention is carried out, for example, in a direct channel of the communication system shown in FIG. 1 in order to increase its throughput.

In systems with a high data rate, as a rule, they use a very wide frequency band. Under these conditions, the MIMO channel experiences distortions of frequency selectivity, which in the time domain appear as multipath. An effective method of combating multipath is orthogonal frequency division multiplexing (OFDM), which is equivalent to representing a single frequency-selective channel by a plurality of frequency subchannels in which frequency selectivity is absent. This fact is reflected in the developing standards of modern communication systems such as IEEE 802.16, 802.20, which provides all the basic mechanisms for using MIMO-OFDM technology.

The implementation of the method of the claimed invention will be discussed below on the example of a MIMO-OFDM communication system.

In OFDM systems, the transmission-reception of signals between transmitting and receiving stations is carried out through many physical communication channels. Orthogonal frequency subchannels corresponding to different OFDM signal subcarriers are used as physical channel data.

The increase in throughput by the method of the claimed invention is realized due to the joint servicing of several subscriber stations via the same physical channels.

In order for the signals to not cause mutual interference at the receiving points, joint processing of the AC data signals is performed on the BS immediately before transmission. In this case, information about the communication channel is used.

To implement this method, it is necessary that the total number of antennas of subscriber stations does not exceed the number of antennas of the base station. If, for example, a BS is equipped with 4 antennas, N = 4, then at the same time it is possible to serve 4 speakers with one antenna or 2 speakers, each of which has two antennas, or one speaker with one and one AC with three antennas.

This requirement limits the number of simultaneously served subscribers. However, there can be a fairly large number of users in a communication system. Therefore, in the control unit 4 of the base station 1, the organization of AC 6.1-6.L is performed by combining them into groups. In the process of organization, AS groups of joint and individual services are determined. In each joint service group, all speakers serve together in the frequency subchannels common to the speakers in this group. In the group of individual service ACs, each AC is served individually by means of frequency subchannels allocated only to this AC. In the process of grouping, various parameters are used, for example, the number of antennas, the degree of loading of the base station, long-term information about the communication channel, the channel quality indicator of each speaker, etc.

The method of the claimed invention can be used in a joint service group of speakers, as well as for individual service of those speakers that have more than one antenna.

Consider in more detail the implementation of the claimed invention with reference to Fig.2-7.

In the structural diagrams (Fig.2-7) devices and synchronization signals are not shown, although it is understood that they are necessary and must be present when implementing the blocks that are part of the structural diagrams of devices on which the inventive method is carried out. The synchronization of signals in devices is performed by any known method (traditionally) for these communication systems and does not change with respect to the algorithm according to the claimed method, therefore, for simplicity of presentation, the description of synchronization signals and, accordingly, devices that perform these functions is omitted.

Also, for simplicity, all connections (connections) in the structural diagrams are shown by lines of the same thickness (bus), despite the fact that some connections reflect the transmission of single digital and analog signals, while others reflect the transmission of arrays of signal-matrices, since all these signals have a complex structure .

Figure 2 is a structural diagram of the transmitter 2 of the communication system MIMO-OFDM, which carry out the method according to the claimed invention.

The transmitter 2 (figure 2) contains:

signal conditioners for joint service groups 11.1-11.U, where U is the maximum number of joint service groups,

signal shaper for individual service speakers 12,

overhead driver 13,

OFDM modulators 14.1-14.N.

The first K inputs of each of the formers 11.1-11.U receive information messages intended for transmission to subscriber stations of the corresponding group. The second F inputs of each of the formers 11.1-11.U receive channel matrix estimates for those subcarriers that are used to serve the speakers of the corresponding group. At each of the N outputs of each of the U shapers, F subcarrier signals are generated for transmission through a respective transmit antenna.

The number and numbers of used subcarriers come as control signals from the BS control unit through the control inputs of the shapers 11.1-11.U.

For reasons of readability, the control signals are not shown in the structural diagrams (Figs. 2-5), although it is understood that they come from the BS control unit to the control inputs of the blocks included in the structural diagrams of devices on which the inventive method is implemented.

The signals intended for transmission formed at the outputs of the shapers 11.1-11.U are fed to the inputs of OFDM modulators 14.1-14.N.

The signals for the individual service ACs are generated in the shaper 12. In this case, frequency channels (subcarriers) are determined in the BS control unit 4 for communication with each of the ACs of this group. When generating signals, these subcarriers are taken into account, as well as modulation types, coding methods, and transmission-reception methods provided for by the used communication standard (see, for example, [2]) and defined in the system for individual service subscriber stations.

The signals intended for transmitting personal service speakers generated at the outputs of the driver 12 are fed to other inputs of OFDM modulators 14.1-14.N.

In the service signal generator 13, service signals necessary for implementing communication in the MIMO-OFDM system are generated, such as pilot signals, null carrier signals, guard bands, DC carriers , [2] 802.16TM IEEE Standard for tocal and metropolitan area networks. Part 16: Air Interface for Fixed Broadband Wireless Access Systems, October 1, 2004.

The signals generated in the shapers 12 and 13 are fed to other inputs of OFDM modulators. Thus, the inputs of each of the OFDM modulators 14.1-14.N receive the signals of all frequency subchannels used in the system, intended for transmission through a transmitting antenna connected to the output of this OFDM modulator. OFDM modulators 14.1-14.N perform typical operations for generating an OFDM signal (inverse discrete Fourier transform, adding a cyclic prefix - see, for example, [9] John G.Proakis, "Digital Communication," McGrow-Hill, Third Edition) as well as conversion to analog form, transfer to the radio frequency region and signal processing at the radio frequency. After that, the generated radio signals are transmitted through transmitting antennas 15.1-15.N.

For a better understanding of the implementation of the invention, let us consider in more detail the operation of the signal shaper of the joint service group AC 11.1-11.U, the structural diagram of one of which is made in Fig. 3 and shown as an example of implementation. Each signal shaper of the joint service group AS 11.1-11.U consists of a node for generating information packets 16 and F of the multi-user processing nodes 17.1-17.F, where F is the maximum number of frequency subchannels allocated for communication with subscribers of the joint service group AS.

From the first To the inputs of the signal shaper of the joint service group AC 11, the informational messages intended for transmission to the subscriber stations of this subscriber group are respectively sent to the inputs of the information packet forming unit 16. At node 16, F sequences of modulation symbol packets are formed from these messages. The sequence generated at the separate output of node 16 is intended for transmission in the corresponding frequency subchannel.

From each of the F outputs of the information packet forming unit 16, the sequence of modulation symbol packets arrives at the first input of the corresponding multi-user processing node 17.1-17.F. At the second input of each multi-user processing node, an estimate of the channel matrix of the corresponding frequency subchannel is received from the second inputs of the signal shaper of the joint-service AC group 11.

At each nth output (n = 1, ... N) of each f-th node (f = 1, ... F) of multi-user processing, a signal is generated for transmission through the nth transmit antenna in the f-th frequency subchannel.

The set of signals generated in this way at the outputs of the multi-user processing nodes 17.1-17.F is fed to the outputs of the shaper of the joint-service AC group 11. From the outputs of the shapers 11.1-11.U, this set of signals is fed to the corresponding inputs of the corresponding OFDM modulators 14.1-14 .N (figure 2).

Let us consider in more detail the procedure for generating information packets. Figure 4, as an example of the proposed method, presents a structural diagram of the implementation of the node forming information packets 16. The node forming information packets consists of K parallel channels of signal processing. Each channel carries out the processing of a signal from one of the K subscriber stations of the AC joint service group. The signal processing channel of a separate AS is formed by series-connected coding subassembly 18, modulator 19 and distribution subassembly over subchannels 20.

The inputs of the encoding subnodes 18.1-18. K are the inputs of the node forming information packets 16.

To the inputs of the node forming the information packets 16 are received respectively To the sequence of binary characters. These sequences come from the control unit BS 4 (figure 1), where they are formed from information messages intended for transmission to subscriber stations, respectively.

In the coding subassembly 18 of each of the K processing channels, coding and interleaving of the input binary symbol sequence are performed. In the modulator 19, modulation of the obtained coded sequence of binary symbols is performed. Coding, interleaving and modulation operations are performed in accordance with the selected coding and modulation types, as well as interleaving algorithms provided for by the communication standard used (see, for example, [2] 802.16ТМ IEEE Standard for local and metropolitan area networks. Part 16: Air Interface for Fixed Broadband Wireless Access Systems, October 1, 2004).

The sequence of modulation symbols generated at the output of the modulator 19, is fed to the input of the distribution subassembly for subchannels 20, where this sequence is distributed between F frequency subchannels. Thus, at the F outputs of the subassembly 20, subsequences of modulation symbols are formed for transmission in the respective frequency subchannels.

From the outputs of the subnodes 20.1-20. K, the generated subsequences are fed to the inputs F of the shapers of the frequency subchannel packets 21.1-21.F, so that the subsequences of different subscribers, intended for transmission in the same frequency subchannel, are fed to the inputs of the shaper 21.1-21 .F of the corresponding frequency subchannel.

In each of the F shapers 21.1-21.F, a sequence of modulation symbol packets is formed for transmission to subscriber stations of the served group in the corresponding frequency subchannel.

These packets are formed in such a way that each packet contains M modulation symbols, where M is the total number of receiving antennas of this group of subscriber stations. Moreover, the number of characters intended for transmission of each speaker corresponds to the number of its receiving antennas. Each packet generated at the output of one of the formers 21.1-21.F, in the course of further processing, is represented as an M-dimensional vector of transmitted modulation symbols a = [a ₁ , ... a _M ] ^T.

The generated sequence of modulation symbol packets intended for transmission to subscriber stations of the served group in the corresponding frequency subchannels from the outputs F of the shapers 21.1-21.F are supplied to the outputs of the information packet forming unit 16.

Let us consider in more detail the procedure of multi-user processing, which is performed by means of nodes 17.1-17.F. The structural diagram of the node 17, as an example implementation, is presented in figure 5. Each of the multi-user processing nodes 17.1-17.F consists of a signal vector conversion sub-node 22, a channel matrix processing sub-node 23, a summing sub-node 24, a disturbance vector generating sub-node 25, a linear transformation sub-node 26, and a normalization sub-node 27, with the first input of the multi-user processing node 17 is the input of the signal vector conversion sub-node 22, and the second input is the input of the channel matrix generation sub-node 23, the output of the signal vector conversion sub-node 22 is connected to the first inputs of the sub-node and summing 24 and the sub-node generating the perturbation vector 25, the first output of the sub-processing node of the channel matrix 23 is connected to the second input of the sub-node generating the perturbing vector 25, the third input of which is combined with the second input of the sub-node of the linear transformation 26 and connected to the second output of the sub-processing node of the channel matrix 23 the sub-node of the formation of the disturbing vector 25 is connected to the second input of the summing sub-node 24, the output of which is connected to the first input of the linear conversion sub-node 26, the N outputs of which are connected with N inputs of the normalization subassembly 27, N outputs of which are outputs of the multi-user processing node 17.

The multi-user processing node 17 operates as follows.

From the first input of the node 17 to the input of the signal vector conversion subassembly 22, a sequence of vectors of transmitted modulation symbols a of one of the frequency subchannels is generated at the corresponding output of the information packet forming unit. In subnode 22, the M-dimensional vector of the transmitted modulation symbols a is converted into a 2M-dimensional real-valued vector of the transmitted modulation symbols a _r in accordance with the formula

where Rea, Ima are vectors composed of real and respectively imaginary parts of the corresponding elements of vector a. The real-valued vector a _r thus formed from the output of the subassembly 22 arrives simultaneously at the first inputs of the subassemblies of the summation 24 and the formation of the disturbing vector 25.

The input channel processing node of the channel matrix 23 from the second input of the node 17 receives the channel matrix H of the corresponding frequency subchannel.

Each element of the channel matrix h _{j, i} is an estimate of the signal transmission coefficient over the spatial channel formed by the ith BS transmitting antenna and the jth receiving antenna of this group of subscriber stations. This transmission coefficient is usually represented as a complex number whose module reflects the change in amplitude, and the argument is the change in the phase of the signal when passing through the corresponding spatial communication channel.

There are various ways to obtain estimates of these coefficients. For example, if the communication system uses the time separation of the forward and reverse channels, then the estimation data is generated at the base station from the signals of the reverse channel received from subscriber stations. If the communication system uses the frequency separation of the forward and reverse channels, then the estimates of the channel matrix elements are generated in the receivers of the subscriber stations 8.1-8.L and transmitted to BS 1 via the feedback channel. Methods for channel estimation in a MIMO-OFDM system are known from the literature, for example, [12] Z. Jane Wang, Zhu Han, and KJRay Liu, "A MIMO-OFDM Channel Estimation Approach Using Time of Arrivals," IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, Vol. 4, NO.3, MAY 2005, pp. 1207-1213.

In the conversion node of the channel matrix 23, the following algorithm is implemented.

1. The channel matrix H having the dimension M × N is converted into a real-valued matrix H _{r of} dimension 2M × 2N in accordance with the formula

where ReH, ImH are matrices composed of real and respectively imaginary parts of the corresponding elements of the matrix N.

2. From a real-valued channel matrix H _r form a matrix of preliminary linear signal conversion, according to the formula

where (.) ^H is the symbol for transposition and complex conjugation,

(.) ^-1 - matrix inversion symbol.

The matrix W _r thus formed is fed to the first output of the subassembly 23 and then to the second input of the linear transformation subassembly 26 and the third input of the subassembly of the formation of the disturbing vector 25.

3. By reducing the lattice basis, the matrices W _r form an integer matrix T with a determinant of ± 1, multiplication by which converts the preliminary linear transformation matrix into the matrix Z = W _r T, which has a knowingly low condition number.

They use, for example, the well-known lattice-base reduction algorithm, named after the abbreviation of the authors' names LLL (Lenstra-Lenstra-Lovasz) and presented in [10] by Dirk Wubben, Ronald Böhnke, Volker Kühn, and Karl-Dirk Kammeyer, "Near-Maximum -Likelihood Detection of MIMO Systems using MMSE-Based Lattice Reduction ", IEEE Proc. International Conference on Communications (ICC), Paris, France, June 2004 and [11] A.K. Lenstra, H.W. Lenstra, and L. Lovasz, "Factoring potynomials with rational coefficients", Mathematische Annaten, vol. 261, pp. 515-534, 1982.

Thus formed, the matrix T from the output of the subassembly 23 enters the second input of the subassembly of the formation of the disturbing vector 25.

The subunit of the formation of the disturbing vector 25 operates in accordance with the following algorithm.

1. Using the matrix T, and a real-valued vector of the transmitted modulation symbols a _r , determine the non-quantized perturbation vector as

where A is a real number such that the real and imaginary part of any modulation symbol α in absolute value is strictly less than A / 2, i.e.

2) round off each of the elements of the resulting vector z to the nearest integer in magnitude, thus determining the first quantized vector z ₁ , and at the same time, determine the vector d _{1 of the} corresponding quantization error values by the formulas

where Q (z) is the operation of elementwise rounding of the vector z to the nearest integer,

3) form the second quantized vector z ₂ by determining for each element of the vector z the second closest integer with the opposite value of the quantization error, as well as the second vector of the corresponding values of the quantization error d ₂ according to the formulas

where sign (a) denotes the vector obtained from the vector a by applying the operation to each of its elements

4) from the elements of the first z ₁ and second z ₂ quantized vectors, R quantized vectors u are formed having the smallest values of the total quantization error of the vector, and the total quantization error of the vector is defined as the sum (or sum of squares) of the absolute quantization errors of all elements of the vector,

5) each of R quantized vectors u is transformed by the formula

thus forming many candidate disturbing vectors,

минимальна.6) determine the optimal perturbing vector p ₀ as a vector from the set of candidate perturbing vectors for which the decisive function

is minimal.

The optimal perturbing vector p ₀ thus formed from the output of the subassembly 25 is fed to the second input of the summing subassembly 24, where the real-valued vector of the transmitted symbols and the optimal perturbing vector are summed, thus forming a perturbed real-valued vector of the transmitted symbols (a _r + p ₀ ).

The perturbed real-valued vector of the transmitted symbols (a _r + p ₀ ) from the output of the sub-node 24 is fed to the first input of the sub-node of the linear transformation 26.

In subnode 26, a preliminary linear transformation of the resulting perturbed real-valued vector of modulation symbols is performed, forming a real-valued vector of transmitted signals in accordance with the formula:

From the obtained real-valued vector of transmitted signals x _r in subnode 26 form an unnormalized vector of transmitted signals

where j is the imaginary unit, and x _r (n: m) denotes a vector composed of a sequence of elements of the vector x _r from the n-th to the m-th.

Thus, at the N outputs of the linear conversion subassembly 26, respectively, N elements of an irregularized vector of transmitted signals x _{0 are formed} . From the outputs of the subassembly 26, these signals are fed to the corresponding inputs of the normalization subassembly 27, where they form the vector of the transmitted signals of a given frequency subchannel by multiplying the unnormalized transmitted signals by the normalization coefficient C _T ,

The normalization coefficient C _T is a real number, which is chosen so that the average transmit power of the signals of the generated vector x is equal to the power of the signals transmitted to the receiving stations without multi-user conversion.

The set of signals corresponding to the elements of the received vector x is fed to the outputs of the normalization subassembly 27 and, accordingly, to the outputs of the multi-user processing node 17. The set of signals corresponding to the elements of the obtained vector x are transmitted in the corresponding frequency subchannel through all transmitting antennas - one signal through the antenna. For this purpose, the signals intended for simultaneous transmission through each n-th antenna (n = 1, ... N) in the frequency subchannels allocated for signal transmission to the joint-service speaker groups are supplied to the corresponding inputs of the nth OFDM modulator.

The functions of OFDM modulators 14.1-14.N are presented above when describing the operation of transmitter 2.

Let us consider in more detail the implementation of the claimed invention in the receiver of the speaker, the structural diagram of which is made in Fig.6.

The receiver of the subscriber station contains M _k receiving antennas 7.1-7.M _k and the same number of signal processing blocks of the receiving antennas 28.1-28.M _k and the decoding unit 29, where k is the subscriber station index.

The signal from each receiving antenna 7.1-7.M _{k is} fed to the input of the corresponding signal processing unit 28.1-28.M _k . In each of the blocks 28.1-28.M _k , signal processing is performed, the result of which is F sequences of binary symbols received by the corresponding receiving antenna in F frequency subchannels.

Each signal processing unit of the receiving antenna 28.1-28.M _k (a block diagram as an example of implementation is performed in Fig. 7) contains an OFDM 30 demodulator, F of normalization nodes 31.1-31.F and the same number of modular reduction nodes 32.1-32.F and demodulators 33.1-33.F.

The input of the OFDM demodulator is the input of the signal processing unit of the receiving antenna 28.F. The outputs F of demodulators 33.1-33.F are simultaneously F outputs of the signal processing unit of the receiving antenna.

The OFDM 30 demodulator performs signal processing at the radio frequency, synchronization of the OFDM signal, removal of the cyclic prefix and discrete Fourier transform, the result of which are the signals F of the frequency subchannels. OFDM signal demodulation operations are presented, for example, in [9] by John G. Proakis, "Digital Communication", McGrow-Hill, Third Edition.

Thus, on each of the F outputs of the OFDM 30 demodulator, a signal y is generated, which is a complex number with a module and an argument corresponding to the amplitude and phase of the signal received in this frequency subchannel.

The signal of each of the F frequency subchannels is processed independently in the corresponding processing subchannel formed by series-connected normalization unit 31, modular reduction unit 32, and demodulator 33.

In each processing sub-channel of each receiving antenna, the following operations are performed.

In the normalization unit 31, the signal y is normalized, multiplying it by the normalization coefficient C _R , thus forming a normalized signal

where the normalization coefficient C _{R is} set, for example, equal to the reciprocal of the normalization coefficient of the transmission:

In the node modular reduction 32 determine the real and imaginary parts of the normalized signal y _norm

With the received signals z and c in node 32, the operation of nonlinear modular reduction modulo A is performed:

where [x] is the integer part of x, that is, the maximum integer less than x.

и

формируют комплексный сигнал

который поступает на выход узла модульного редуцирования 32 и далее на вход демодулятора 33.From signals

and

form a complex signal

which is fed to the output of the modular reduction unit 32 and then to the input of the demodulator 33.

обычным способом, формируя последовательность оценок принятых двоичных символов.In demodulator 33, complex signal demodulation is performed

in the usual way, forming a sequence of estimates of the received binary characters.

The thus formed sequences of estimates of binary symbols adopted in the F frequency subchannels from the outputs of the demodulators 33.1-33.F are fed to the outputs of the signal processing unit of the receiving antenna 28 and then to the corresponding inputs of the decoding unit 29.

Thus (see Fig. 6), the generated sequences with F outputs of each of the blocks 28.1-28.M _k are supplied to the corresponding inputs of the decoding unit 29, where F sequences of binary symbol estimates are combined and deinterleaving and decoding operations are performed inverse to those which were used in one of the coding subnodes 18.1-18. K of the BS transmitter (Fig. 4) corresponding to this k-th subscriber station.

Thus, at the output of the decoding unit 29 of the receiver of the subscriber station 8, a sequence of binary symbols of the received message is formed.

To assess the noise immunity characteristics of the signal transmission-reception algorithm developed in accordance with the method according to the claimed invention, computer simulation was performed.

For simulation, a software model of the transmitter was developed with N = 4 antennas and 4 receivers of subscriber stations, each of which is equipped with one receiving antenna.

The developed model corresponds to one group of joint service stations. The structure of such a model is described above and shown in figures 1-7.

For simplicity, only one physical (frequency) channel was used in the program model for transmitting and receiving a signal in a multi-user MIMO communication system.

The simulation results of multi-user MIMO algorithms with a total spectral efficiency of 8 bits / sec / Hz are shown in Fig. 8. The curves in this drawing reflect the dependence of BER on E _B / N ₀ , where BER is (bit error rate) the probability of receiving a signal bit averaged over all subscriber stations, E _B / N ₀ is the average ratio of the energy of the signal bit E _B to spectral density noise power N ₀ at the receiving point. It is assumed that the reception conditions and data rates are the same for all speakers.

When modeling the algorithms, convolutional coding was used with a rate of 1/2 and the size of the original block of uncoded binary characters of 192 bits. A channel model with block fading and additive Gaussian noise was used.

Algorithm Modeling

- pre-coding MMSE,

- prototype algorithm and

- algorithm of the proposed solution

made for 4 subscriber stations, each of which has 1 receiving antenna. To transmit the signal of each speaker, 16QAM modulation was used.

The modeling of the precoding algorithm with block diagonalization was performed for 2 subscriber stations, each of which has 2 receiving antennas. Transmission and reception in the generated orthogonal MIMO channels of each speaker was performed according to the method of its own subchannels. In this case, 2 own subchannels with modulation 64QAM and BPSK were used, respectively.

The characteristics presented indicate that in the working range of BER values (BER <= 0.05), an algorithm that implements the inventive method has maximum noise immunity relative to other considered algorithms.

Thus, the inventive method of transmitting and receiving a signal in a multi-user radio communication system with multiple transmitting and multiple receiving antennas can significantly increase the throughput of the multi-user MIMO communication system, since it performs simultaneous servicing of a group of several subscriber stations in the same physical channel.

In terms of noise immunity, the algorithm that implements the claimed method surpasses the well-known multi-user MIMO algorithms [4] QHSpencer, and M.Haardt, "Capacity and Downlink Transmission Algorithms for a Multi-user MIMO Channel", Signals, Systems and Computers, 2002. Conference Record of the Thirty-Sixth Asilomar Conference, Volume 2, Issue, 3-6 Nov. 2002 Page (s): 1384-1388 vol.2, [5] JC Mundarath, JHKotecha, "Zero-Forcing Beamforming for Non-Collaborative Space Division Multiple Access", Proceedings of 2006 IEEE International Conference on Acoustics, Speech and Signal Processing ICASSP, May 14-19, 2006, Volume: 4, page (s): IV-IV, [6] A Wicsel, YCEtdar, and Sh.Shamai, "Optimal Generalized Inverses for Zero Forcing Preceding", 41st Annual Conference on Information Sciences and Systems, CISS '07, 14-16 March 2007, pages: 130-134, and also the algorithm that served as a prototype of the claimed method [8] Christoph Windpassinger, Robert FH Fischer, and Johannes B. Huber, "Lattice-Reduction-Aided Broadcast Preceding ", IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 52, NO.12, DECEMBER 2004, pp.2057-2060.

This advantage of the proposed method is achieved by using a new sequence of interrelated actions, including the vector perturbation procedure in combination with the reduction of the lattice basis and multi-alternative quantization.

In addition, an advantage of the method according to the claimed invention is the possibility of a relatively simple implementation in the transmitter and, especially, a simple implementation of the receiver of the subscriber station. In this case, the AS receiver is implemented as independent channels for processing signals of various receiving antennas.

A significant advantage of this invention is that it is possible to implement if there is only one receiving antenna at each of the subscriber stations of the communication system.

Another important advantage of the claimed invention is the possibility of its implementation in almost any distribution environment. Let us explain that for the implementation of traditional single-user MIMO methods, a propagation medium with a large number of scattering objects is necessary, which is not always implemented in practice. At the same time, the method according to the claimed invention provides a gain in throughput even in a relatively low scattering environment, since the receiving side antennas belong to different user terminals, as a result of which their signals have a low correlation regardless of the properties of the signal propagation medium.

Literature

1. G.J. Foshini, G. D. Golden, R.A. Valenzuela, "Simplified processing for high spectral efficiency wireless communication employing multi-element arrays", IEEE Selected Areas Communication, vol. 17, pp. 1841-1852, November, 1999.

2. 802.16TM IEEE Standard for local and metropolitan area networks. Part 16: Air Interface for Fixed Broadband Wireless Access Systems, October 1, 2004.

3. M. Airy, A. Forestnza, R. W. Heath, Jr. S. Shakkottai, "Practical Costa precoding for the multiple antenna broadcast channel", IEEE Global Telecommunications Conference, GLOBECOM, 29 Nov.-3 Dec. 2004, Volume 6, Page (s): 3942-3946.

4. QHSpencer, and M. Haardt, "Capacity and Downlink Transmission Algorithms for a Multi-user MIMO Channel", Signals, Systems and Computers, 2002. Conference Record of the Thirty-Sixth Asilomar Conference, Volume 2, Issue, 3- 6 Nov. 2002 Page (s): 1384-1388 vol. 2.

5. JC Mundarath, JHKotecha, "Zero-Forcing Beamforming for Non-Collaborative Space Division Multiple Access", Proceedings of 2006 IEEE International Conference on Acoustics, Speech and Signal Processing ICASSP, 14-19 May 2006, Volume: 4, page ( s): IV-IV.

6. A Wiesel, Y. C. Eldar, and Sh. Shamai, "Optimal Generalized Inverses for Zero Forcing Preceding," 41st Annual Conference on Information Sciences and Systems, CISS '07, March 14-16, 2007, pages: 130-134.

7. R.F.H. Fischer, C. Windpassinger, A. Lampe, J.B. Huber, "Space-Time Transmission using Tomlinson-Harashima Precoding", In Proc. 4th Int. ITG Conf., Pp. 139-147, Berlin, Jan. 2002.

8. Christoph Windpassinger, Robert F.H. Fischer, and Johannes B. Huber, "Lattice-Reduction-Aided Broadcast Precoding", IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 52, NO.12, DECEMBER 2004, pp.2057-2060.

9. John G. Proakis, "Digital Communication," McGrow-Hill, Third Edition.

10. Dirk Wubben, Ronald Böhnke, Volker Kühn, and Karl-Dirk Kammeyer, "Near-Maximum-Likelihood Detection of MIMO Systems using MMSE-Based Lattice Reduction", IEEE Proc. International Conference on Communications (ICC), Paris, France, June 2004.

11. A.K. Lenstra, H.W. Lenstra, and L. Lovasz, "Factoring potynomials with rational coefficients", Mathematische Annalen, vol. 261, pp. 515-534, 1982.

12. Z. Jane Wang, Zhu Han, and K.J. Ray Liu, "A MIMO-OFDM Channel Estimation Approach Using Time of Arrivals", IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 4, NO.3, MAY 2005, pp. 1207-1213.

Claims (6)

1. The method of transmission-reception of a signal in a multi-user radio communication system with multiple transmitting and multiple receiving antennas, which use a transmitting station equipped with N transmitting antennas, and K receiving stations, where K≥2, and each receiving station is equipped with at least one receiving antenna, and the total number of receiving antennas of the receiving stations M satisfies the condition 1 <M≤N, while the transmission-reception of signals between the transmitting and receiving stations is carried out through F physical communication channels Where F≥1, consists in the fact that for each of the physical channels F estimate spatial parameters plurality of communication channels, each of which is formed by one transmit antenna of a transmitting station and a receiving antenna of the receiving station; transmit and receive signals between the transmitting station and receiving stations using F physical channels, for which K sets of modulation symbols are formed at the transmitting station for transmission to K receiving stations, respectively, F packets of M modulation symbols are formed from K formed sets of modulation symbols each, including in a packet of M _k modulation symbols for each k-th receiving station, where M _k is the number of receiving antennas of the k-th receiving station; each of the F modulation symbol packets is transmitted over the corresponding physical channel, wherein the modulation symbol packet is represented as a vector of transmitted modulation symbols a = [α ₁ ... α _M ] ^T , each element of which is a complex number, with a module and an argument that reflect the amplitude, and accordingly the phase of the corresponding modulation symbol, perform multi-user conversion of the vector of transmitted modulation symbols a into the vector of transmitted signals x so that the transmitted signals ala do not cause interference in the M receiving antennas of receiving stations, which form the channel matrix H for the physical channel using spatial transmission channels coupling coefficients of the vector of transmitted modulation symbols a and a channel matrix H formed real-valued vector a _r, and the matrix H _r in accordance with the formulas

where ReY, ImY are matrices composed of real and respectively imaginary parts of the corresponding elements of the matrix Y,
from a real-valued channel matrix H _r form a matrix W _{r of} preliminary linear signal transformation, by reducing the basis of the lattice matrix W _r form an integer matrix T with a determinant of ± 1, multiplication by which converts the matrix of preliminary linear transformation into a matrix Z = W _r T having a knowingly low condition number using the matrix T and a real-valued vector of transmitted modulation symbols a _r , determine the non-quantized perturbation vector, as

where A is a real number, and such that the real and imaginary part of any modulation symbol in absolute value is less than A / 2, round off each of the elements of the resulting vector z to the nearest integer in magnitude, thus determining the first quantized vector z ₁ , and determine the vector of the corresponding values of the quantization error
z ₁ = Q (z), d ₁ = z ₁ -z,
where Q (z) is the vector obtained by elementwise rounding of the vector z to the nearest integer,
form the second quantized vector z ₂ by determining for each element of the vector z the second closest integer with the opposite quantization error value, as well as the second vector of the corresponding quantization error values, from the elements of the first z ₁ and second z ₂ quantized vectors form K quantized vectors and having the smallest values of the total quantization error of the vector, the total quantization error of the vector is determined, for example, as the sum or the sum of the squares of the absolute quantization errors of all elements ntov vector, each of R u quantized vectors converted by the formula
p = -A · T · u,
thus forming the set of candidate disturbing vectors, the optimal disturbing vector p ₀ is determined as the vector from the set of candidate disturbing vectors for which the decisive function F (p) = || W _r · a _r -p || ^{2 is} minimal, they form a perturbed real-valued vector of transmitted modulation symbols by summing the real-valued vector of transmitted modulation symbols and the optimal perturbation vector and perform preliminary linear transformation of the resulting perturbed real-valued vector of modulation symbols, thus forming a real-valued vector of transmitted signals ,
x _r = W _r (a _r + p ₀ ),
from the received real-valued vector of the transmitted signals x _r form an unnormalized vector of transmitted signals
x ₀ = x _r (1: N) + j · x _r (N + 1: 2N),
where j is the imaginary unit,
and x _r (n: m) denotes a vector composed of a sequence of elements of the vector x _r from n-th to m-th;
form a vector of transmitted signals, multiplying the vector of unnormalized transmitted signals by a normalization coefficient C _T ,
x = x ₀ · C _T
the set of signals corresponding to the elements of the obtained vector x is transmitted in the corresponding physical channel through all the transmitting antennas - one signal through the antenna; receive signals at each of the K receiving stations, and in each physical channel of each receiving antenna, the reception is carried out in such a way that they form the signal y, as a complex number with the module and argument corresponding to the amplitude and phase of the signal received by this physical channel, normalize the signal y, multiplying its normalization coefficient C _R , thus forming a normalized signal
at _norm = y · C _R ,
determine the real and imaginary parts of the normalized signal in _norm
z = Rey _norm , c = Imy _norm , the received signals z and c perform the operation of modular reduction modulo A:

Where

is the integer part of x, that is, the maximum integer less than x,
from signals

and

form a complex signal

using complex signal values

thus formed in each physical channel of each receiving antenna, demodulate and decode the received signal. 2. The method according to claim 1, characterized in that for each of the F physical channels, the transmission coefficient of the communication channel and the signal-to-noise ratio in the channel are used as estimated parameters. 3. The method according to claim 1, characterized in that at the transmitting station, each of the K information messages intended for transmission to the receiving stations is respectively represented as a sequence of binary symbols, and then the encoding, interleaving and modulation of the binary symbols of this sequence are performed. 4. The method according to claim 1, characterized in that the matrix W _r preliminary linear signal conversion form as

where H _r is a real-valued channel matrix for a given physical channel. 5. The method according to claim 1, characterized in that the second quantized vector z ₂ and the second vector of the corresponding quantization error values are formed by the formulas
z ₂ = z ₁ -sign (d ₁ ), d ₂ = z ₂ -z,
where sign (a) denotes the vector obtained from the vector a,

applying operations to each of its elements. 6. The method according to claim 1, characterized in that when forming the vector of the transmitted signals, the normalization coefficient C _{T is} selected so that the average transmit power of the signals of the generated vector x is equal to the power of the signals transmitted to the receiving stations without multi-user conversion.

Images