RU2350025C2 - Method of reception of multicomponent signal in radio communication system with n transmission channels and m channels of reception (versions) and device for its realisation (versions) - Google Patents

Abstract

FIELD: information technologies.

SUBSTANCE: instead of traditional quantisation and formation of estimates of figures of modulation from the gained quantised estimates, carry out the multialternative (multiple) quantisation and formation of candidate vectors set. A candidate vector is a plurality of figures of the modulation which are quantised (rigid) estimates N of transmitted information signals. The set of candidate vectors is made of several most probable quantised estimates of each of transmitted figures of modulation. Candidate vectors are shaped in space of the modified signals, and then will be conversed to space of initial signals. The gained set of candidate vectors is used for search of the solution minimising decision function of the peak credibility, or for formation of metrics for the soft decoding.

EFFECT: increase of noise stability of a multicomponent signal reception, in system of a radio communication without coding, and in system of a radio communication with coding.

21 cl, 9 dwg

Description

The group of inventions relates to the field of radio engineering, in particular to a method (options) and a device (options) for receiving a multi-component signal in a radio communication system with N transmission channels and M reception channels, and can be used in a communication system with multidimensional signals, for example, in a MIMO communication system .

The problem of receiving multicomponent signals arises in a large number of telecommunications fields, including such as multi-user detection, reception using adaptive antenna arrays, Multiple-Input-Multiple-Output (MIMO) technology, hereinafter referred to as MIMO. The invention will be described by the example of a MIMO communication system.

MIMO technology attracts much attention from developers and researchers of wireless communication systems, due to the possibility of a significant increase in throughput.

In MIMO systems, more than one antenna is used on the transmitting and receiving sides, as a result of which the spatial communication channel between the transmitter and the receiver has multiple inputs — signal transmission channels — and multiple outputs — signal reception channels. The whole set of signal propagation channels between transmitting and receiving antennas is called a MIMO channel. The increase in throughput is achieved due to the simultaneous transmission of various information signals on various spatial subchannels of the MIMO channel.

The instantaneous implementation of the MIMO channel is evaluated at the receiving side and used when receiving a signal. The use of this estimate in signal transmission gives a great advantage to the MIMO system in terms of the ability to minimize the level of mutual interference at the receiving point. The MIMO method, which uses channel estimation for signal transmission, is called closed-loop MIMO (MIMO). However, a more practical approach is open loop MIMO when channel estimation is not available to the transmitting side.

In feedbackless MIMO systems, due to the lack of information about the MIMO channel on the transmitting side, it is not possible to mutually optimize the transmission of information on parallel spatial channels, as a result of which the signals transmitted through various antennas create mutual interference on the receiving side. The most common transmission method in open loop MIMO systems is spatial multiplexing, first proposed as part of the V-BLAST method (see 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 [1], GJ Foschini, GD Golden," Wireless communications system having a space-time architecture employing multi-element antennas at both the transmitter and receiver (Patent style), "US patent: US 6317466 B1. Nov. 13, 2001 [2], GJFoschini," Wireless communications system having a layered space-time architecture employing multi-element antennas, "US Patent US # 6097771, August 1, 2000 [3]).

According to the V-BLAST spatial multiplexing method, the transmitter performs coding, interleaving, and modulation of the stream of binary symbols of the original information message, thereby forming a stream of modulation symbols, each of which represents L consecutive bits of the original stream and belongs to the set m = 2 ^{L of} all possible values.

The modulation symbols s are usually represented by complex numbers, the modulus and argument of each of which represents the amplitude and, accordingly, the phase of the corresponding modulation symbol. The real and imaginary parts of this complex number are the quadrature components of the modulation symbol.

The generated modulation symbol stream is divided into packets, each of which consists of N symbols, where N is the number of transmitting antennas. Radio symbols are formed from the symbols of each packet, forming the corresponding packet of information signals, which are transmitted simultaneously - one signal through each antenna.

Reception is carried out using M≥N receive antennas. The signal of each receiving antenna can be represented by a complex number x, the module of which corresponds to the amplitude, and the argument to the phase of this signal. The set of signals M receiving antennas is usually represented by an M-dimensional vector of complex numbers, which can be expressed as follows

where x = [x ₁ , ... x _M ] ^T is the vector of received signals, s = [s ₁ , ... s _N ] ^T is the vector of modulation symbols corresponding to the transmitted packet of information signals, H is the channel matrix, each element h _{i, j} , which is the normalized coefficient of signal transmission from the jth transmitting to the ith receiving antenna, n = [n ₁ , ... n _M ] ^T is the interference vector of the receiving antennas, which are usually approximated as independent implementations of additive Gaussian noise, [. ] ^T is the transpose sign.

When receiving such a multidimensional signal, the channel matrix H is first estimated and then, using this estimate, the symbols of the vector s are demodulated.

The most effective algorithm for receiving multidimensional signals is the maximum likelihood algorithm (MLA), presented, for example, in J. Jalden, Bjorn Ottersten, "On the Complexity of Sphere Decoding in Digital Communication," IEEE TRANSACTION ON SIGNAL PROCESSING, Vol. 53, No. 4, pp. 1474-1484, April; 2005 [4]. An estimate of the maximum likelihood of a vector of transmitted information symbols s is a value that provides a minimum of the likelihood ratio functional. The evaluation of the transmitted information packet is expressed as

по множеству всевозможных значений вектора s. Сложность реализации данного метода зависит от объема множества поиска, который экспоненциально растет с увеличением количества передающих антенн и количества бит информации, передаваемых через каждую антенну. Поэтому даже при небольшом количестве передающих антенн алгоритм может оказаться весьма сложным при реализации в аппаратуре приемника.where S is the set of all possible values of the vector s . Therefore, the maximum likelihood method involves finding the minimum of the decisive function

by the set of all possible values of the vector s . The complexity of the implementation of this method depends on the volume of the search set, which exponentially increases with the number of transmitting antennas and the number of bits of information transmitted through each antenna. Therefore, even with a small number of transmitting antennas, the algorithm can be very complicated when implemented in receiver equipment.

From a practical point of view, linear reception methods are very attractive in which the estimation of the vector s of the modulation symbols is formed by a linear transformation (see D. Gesbert, "Robust linear MIMO receivers: A minimum error-rate approach," IEEE Trans. Signal Processing, Special issue on MIMO, May 2, 2003 [5]):

where (.) ^t is the symbol of transposition and complex conjugation, W is the matrix of linear transformation coefficients, which is formed in such a way as to optimize the estimate in the sense of some criterion.

The most effective of the linear methods is the minimum mean squared error (MMSE) method. In this case, the matrix W is formed as

where σ ² is the noise variance in each of the receiving antennas, I is the diagonal unit matrix of dimension M × M, (.) ^-1 is the symbol of the inversion of the matrix.

An algorithm is also known that provides zero error in the absence of noise. In foreign publications, this linear algorithm is called the method of zeroing - zero forcing. In this case, the matrix W is formed as

The method of zeroing has a lower noise immunity than MMSE, but it is easier to implement, since it does not require noise dispersion estimation.

According to the noise immunity characteristics, the minimum mean-square error of the estimation method loses to the maximum likelihood method. In the MIMO channel with low conditioning of the matrix H, this loss is negligible. However, the use of the matrix inversion operation by this method leads to the fact that in the MIMO channel with a highly conditioned channel matrix, this loss increases and becomes very significant.

Significant progress in solving the problem of the “bad” conditionality of the MIMO channel matrix was observed using the lattice basis reduction approach (see 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 [6], Christoph Windpassinger, Robert FH Fischer," Low-Complexity Near-Maximum-Likelihood Detection and Precoding for MIMO Systems using Lattice Reduction, "ITW2003, Paris, France, March 31-April 4, 2003 [7], Christoph Windpassinger, Robert FH Fischer," From Lattice-Reduction-Aided Detection Towards Maximum-Likelihood Detection in MIMO Systems, "IEEE Proc. Global Communications Conference (GLOBECOM), Taipei, Taiwan, November 17-21, 2002 [8 ], Huan Yao and Gregory W. Womell, "Lattice-Reduction-Aided Detectors for MIMO Communication Systems," in Proc. IEEE Globecom, Taipei, Taiwan, Nov., 2002 [9].

The idea of this approach is that the channel matrix is represented as the product of a well-conditioned matrix G and a unimodular matrix, that is, a matrix with integer elements and a determinant of ± 1. The demodulation problem is solved with respect to the modified vector of transmitted modulation symbols, which is the product of a unimodular matrix by the initial vector s of modulation symbols.

Demodulation is performed in two stages. At the first stage, suboptimal demodulation of the modified symbol vector is performed using a well-conditioned channel matrix G. Methods based on matrix inversion give high noise immunity at this stage, since a well-conditioned matrix is reversed. The resulting estimate of the modified signal is quantized, due to which noise filtering occurs. At the second stage, the conversion of the modified signal estimate to the estimate of the initial information signal is performed using a unimodular matrix.

The procedure for decomposing a channel matrix into a well-conditioned and unimodular matrix is performed by reducing the lattice basis (see [6] - [9]). Accordingly, this approach to the demodulation of a multidimensional signal is called the lattice reduction method or lattice reduction. It is this approach that underlies the claimed invention.

The implementation of the potentially high throughput of MIMO systems is possible only with the use of efficient noise-resistant coding. Therefore, another important aspect in the design of MIMO signal reception algorithms is that they should be well combined with the error-correcting coding algorithm.

Typically, during reception, the generated symbol estimates are converted to binary form, and fed to a decoder to recover the binary symbols of the original information message. The most efficient decoding is soft decoding. At the same time, soft estimates (decisions) of bits are received at the decoder input - in the form {B-B}, where the sign corresponds to a hard estimate of the transmitted binary character (or bit) 1 or -1, and the absolute value of B is a metric that reflects the probability measure of acceptance by bit given hard value.

The soft decoding metric (or soft decision) of some bit b _k is the log-likelihood ratio (LLR). In the absence of a priori information about the values of the transmitted bits and the mutual independence of the transmitted bits, the logarithm of the likelihood ratio for the kth bit transmitted using the vector s of modulation symbols can be expressed as shown in Dominik Seethaler, Gerald Matz, and Franz Hlawatsch, "Low-Complexity Soft Demodulation of MIMO-BICM Using the Line-Search Detector," Institute of Communications and Radio-Frequency Engineering, Vienna University of Technology. / 0-7803-8939-5 / 05 / (C) 2005 IEEE [10]

и

- множества значений вектора s, для которых k-й бит принимает значение 1 и -1 соответственно,

- норма вектора.Where

and

- the set of values of the vector s for which the kth bit takes the value 1 and -1, respectively,

is the norm of the vector.

для всех элементов множеств

и

. Это значительно усложняет алгоритм приема, так как суммарное количество элементов данных множеств соответствует 2^LN и растет экспоненциально с увеличением числа передающих антенн N и L - количества бит, представляемых одним символом модуляции.The formation of metrics (6) requires the calculation of a function

for all elements of sets

and

. This greatly complicates the reception algorithm, since the total number of elements of these sets corresponds to 2 ^LN and grows exponentially with increasing number of transmitting antennas N and L - the number of bits represented by one modulation symbol.

Linear methods of reception form a soft (not quantized) estimate of each symbol of the vector s and allow simplified formation of soft decisions, which are performed separately for each symbol according to the received estimate of this symbol and the modulation map (see Michael Mao Wang, Weimin Xiao, and Tyier Brown, " Soft Decision Metric Generation for QAM With Channel Estimation Error, "IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 50, NO. 7, Pp.1058-1061, JULY 2002 [11]).

Using the approximation (see [10]), we can write the following expression for the logarithm of the likelihood ratio of the kth bit transmitted using the mth modulation symbol.

- линейная оценка передаваемого символа модуляции s_m,

and

- множества значений символа модуляции s_m, для которого k-й бит принимает значение +1 и -1 соответственно, β_m - коэффициент преобразования символа s_m за счет канала связи и линейного преобразования, формирующего оценку (3).Where

- a linear estimate of the transmitted modulation symbol s _m ,

and

- the set of values of the modulation symbol s _m , for which the kth bit takes the value +1 and -1, respectively, β _m is the symbol conversion coefficient s _m due to the communication channel and the linear transformation forming the estimate (3).

This method of forming soft solutions is much simpler than the method based on expression (6), but it also has a significantly lower noise immunity.

Therefore, another serious problem of receiving a MIMO signal, which is solved in the method of the claimed invention, is the need to develop an effective algorithm for generating soft decisions - metrics for soft decoding.

Closest to the claimed invention is a technical solution described in [6].

The prototype method is as follows.

A known method of receiving a multi-component signal in a radio communication system with N transmission channels and M reception channels (prototype), in which

each of the M components of the multicomponent signal received on the corresponding receive channel is an additive mixture of N information signals, each of which is transmitted on the corresponding transmission channel, and interference,

wherein each of the N information signals corresponds to L binary symbols of the transmitted information message, takes one of m = 2 ^L values corresponding to m modulation symbols, and

is included in each of the M components of a multicomponent signal with a weight coefficient reflecting the transmission coefficient of the spatial channel formed by the transmission channel of this information signal and the reception channel of this component of the multicomponent signal, and it is believed that the transmission coefficients of spatial channels and interference dispersion values of the M reception channels are known or previously evaluated, is that

принимаемого сигнала, таким образом, чтоform a vector

received signal in such a way that

each of the M components of the signal is received, determining for each of them a complex number x , the module of which reflects the amplitude, and the argument is the phase of this component of the multicomponent signal, forming the M-dimensional complex vector x of the received signal from these complex numbers, and transform this vector to vector received signal according to the formula

используя коэффициенты передачи пространственных каналов, таким образом, что между

вектором принимаемого многокомпонентного сигнала и сформированной канальной матрицей выполняется соотношениеform a channel matrix

using spatial channel transmission coefficients in such a way that between

the vector of the received multicomponent signal and the generated channel matrix, the relation

- вектор символов модуляции, соответствующих N информационным сигналам,

- вектор помех М каналов приема,Where

is a vector of modulation symbols corresponding to N information signals,

- interference vector M receiving channels,

for this purpose, from the coefficients h _{i, j} , the transmission of spatial channels, each of which is a complex number, with a module and an argument reflecting the change in the amplitude and phase of the j-th information signal received in the i-th reception channel, respectively, during its distribution the corresponding spatial channel, form a complex channel matrix H , which is converted into a matrix H _r according to the formula

where Re H and Im H are the matrices of the real and imaginary parts of the elements of the matrix H , respectively, the matrix H _{r is} expanded by supplementing the matrix σ I from below, where σ ² is the interference variance of each of the M receiving channels, I is the unit diagonal matrix of dimension 2N × 2N

thus forming a channel matrix

в матрицу G, характеризующуюся заведомо низким числом обусловленности, для чегоtransform the channel matrix

into the matrix G , characterized by a knowingly low condition number, for which

формируют целочисленную матрицу Т с определителем, равным ±1, и формируют модифицированную канальную матрицу G, как произведение

by reducing the matrix lattice basis

form an integer matrix T with a determinant equal to ± 1, and form a modified channel matrix G , as the product

умноженный слева на инверсию матрицы Т, то есть

и оценивают его, используя канальную матрицу G и вектор

принимаемого сигнала, формируя таким образом вектор

оценок модифицированных символов модуляции по формулеdefine a vector z of modified modulation symbols as a vector of modulation symbols

multiplied on the left by the inverse of the matrix T , i.e.

and evaluate it using the channel matrix G and the vector

received signal, thus forming a vector

estimates of modified modulation symbols according to the formula

- вектор принимаемого сигнала, расширенный снизу 2М-мерным вектором нулей 0;where (.) ', (.) ^-1 symbols of transposition and inversion of the matrix, respectively,

- vector of the received signal, expanded from below by a 2M-dimensional vector of zeros 0 ;

оценок модифицированных символов 5 модуляции, определяя для каждого элемента данного вектора квантованное значение, как модифицированный символ модуляции, обеспечивающий наименьшую ошибку квантования, и формируют таким образом вектор u, элементы которого являются квантованными оценками модифицированных символов модуляции;quantize vector

estimates of the modified modulation symbols 5, defining a quantized value for each element of a given vector as a modified modulation symbol that provides the smallest quantization error, and thus form a vector u whose elements are quantized estimates of the modified modulation symbols;

transform the vector u by multiplying on the left by the matrix T

thus forming a vector r , the elements of which are real estimates of the modulation symbols corresponding to N transmitted information signals;

оценок символов модуляции, соответствующих N передаваемым информационным сигналам, по формулеusing the vector r , form the vector

estimates of modulation symbols corresponding to N transmitted information signals according to the formula

- мнимая единица, и,where r 1, r 2 - N-dimensional vectors composed of the first and second half of the elements of the vector r, respectively,

- imaginary unit, and,

принимают решение о передаваемом информационном сообщении.using vector

decide on the transmitted information message.

The implementation of the prototype method is presented in article [6] on the example of a MIMO communication system, the structural diagram of which is made in figure 1.

The communication system (MIMO) in accordance with figure 1 contains a device 1 for transmitting a multi-component signal (transmitter) and device 2 for receiving a multi-component signal (receiver). Moreover, the device 1 for transmitting a multi-component signal comprises a coding unit 3, a demultiplexing unit 4, N modulation blocks 5-1 to 5-N, N channels 6-1 to 6-N for transmitting a multi-component signal, and accordingly N transmit antennas 7-1 to 7- N. The device 2 for receiving a multi-component signal contains M antennas 8-1 to 8-M for receiving a multi-component signal, respectively M channels 9-1 to 9-M for receiving, a channel estimator, a demodulation unit 11, and a decision unit 12.

The prototype for the claimed device is a device for receiving a multi-component signal in a radio communication system (MIMO), the structural diagram of which is made in figure 2, it implements the prototype method.

The prototype device (figure 2) contains M antennas 8 - 1-8-M for receiving a multi-component signal, respectively, M channels 9-1-9-M reception, block 10 channel estimation, block 11 demodulation and block 12 decision, the outputs of the M antennas 8-1 to 8-M for receiving a multi-component signal are connected respectively to the inputs of the M channels 9-1 to 9-M of the reception, the outputs of which are connected via a data bus to the input of the channel estimator 10 and the first input of the demodulation unit 11, the input of the channel estimation block and the first input of the demodulation block are signal inputs, the second and third inputs the demodulation unit are connected respectively to the first and second outputs of the channel estimator 10, which forms at the first output an estimate of the complex transmission coefficients of the signal during propagation from each of the transmitting antennas to each of the receiving antennas, and at the second output, estimates of the noise variance of the receive channels, output 11 the demodulation unit is connected to the input of the decision block 12, the output of which is the output of the device, the demodulation block 11 contains a node 13 for generating a vector of the received signal, a node for converting the channel m tritsy, node 15 for evaluating modified modulation symbols, node 16 for quantization, node 17 for estimating modulation symbols and node 18 for evaluating binary symbols, while the input of node 13 for generating a vector of the received signal is the first input of block 11 of the demodulation, the first and second inputs of node 14 for transforming the channel matrix are respectively the second and third inputs of the demodulation unit 11, the output of the received signal vector forming unit 13 is connected to the first input of the modified modulation symbol evaluation unit 15, the second input of which is is dined with the first output of the channel matrix transform node 14, the output of the modified modulation symbol estimation node 15 is connected to the first input of the quantization node 16, the second input of which is combined with the first input of the modulation symbol estimation node 17 and connected to the second output of the channel matrix transform node 14, the second input the modulation symbol estimator 17 is connected to the output of the quantization node 16, the output of the modulation symbol estimator 17 is connected to the input of the binary symbol estimator 18, the output of which is the output of the demodulation block 11.

Carry out the prototype method on the device (figure 2) as follows.

In the device for receiving a multi-component signal, an input signal is received via M antennas 8-1 to 8-M, the output of each of which is connected to the input of the corresponding reception channel 9-1 to 9-M.

In each of the M channels 9-1 - 9-M receive perform the functions of signal processing at the radio frequency, synchronization, as well as transfer to the region of the video frequency and conversion to digital form.

At the output of each of the M channels 9-1 - 9-M reception, a received signal is formed as a complex number, the module of which reflects the amplitude, and the argument reflects the phase of this signal.

In the process of further signal processing, the set of received signals of the M channels 9-1 - 9-M reception is considered as an M-dimensional complex vector of the received signal x = [x ₁ , ... x _N ] ^T.

The vector x of the received signal generated in this way is fed (via the bus) to the signal inputs — the first input of the demodulation block 11 and the input of the channel estimation block 10.

In block 10 of the channel estimation, the channel matrix H is estimated and the noise variance of the reception channels σ ^{2 is} estimated. At the first output, the channel estimator 10 generates estimates of the complex signal transmission coefficients during propagation from each of the transmitting antennas to each of the receiving antennas, and at the second output, estimates of the noise variance of the receive channels.

The output signals from the first and second outputs of the channel estimation block 10 are respectively supplied to the second and third inputs of the demodulation block 11.

From the first input of the demodulation unit 11, the signal is supplied to the inputs of the node 13 for forming the vector of the received signal. From the second input of the demodulation unit, the signal is supplied to the first input of the channel matrix conversion unit 14. From the third input of the demodulation unit 11, the signal is supplied to the second input of the channel matrix conversion unit 14.

в соответствии с формулой (8).In the node 13 of the formation of the vector of the received signal, the M-dimensional vector x is converted to a 2M-dimensional real vector

in accordance with formula (8).

In the node 14 of the channel matrix transformation, the complex-valued channel matrix H is converted to a real-valued matrix H _r according to the formula

where Re H and Im H are the matrices, respectively, of the real and imaginary parts of the elements of the matrix H , the matrix H _{r is} expanded, supplementing from below with the matrix σ I ,

.where σ ² is the interference variance of each of the M reception channels, I is the unit diagonal matrix of dimension 2N × 2N, thus forming a channel matrix

.

формируют матрицу Т редукции базиса решетки. При этом используют, например, известный алгоритм Lenstra-Lenstra-Lovasz (LLL) (см. [6] и А.К. Lenstra, H.W. Lenstra, and L. Lovasz, "Factoring polynomials with rational coefficients," Mathematische Annalen, vol.261, pp.515-534, 1982 [12]). С использованием матриц

и Т формируют модифицированную канальную матрицу

.By matrix

form the matrix T of the reduction of the basis of the lattice. They use, for example, the well-known Lenstra-Lenstra-Lovasz (LLL) algorithm (see [6] and A.K. Lenstra, HW Lenstra, and L. Lovasz, "Factoring polynomials with rational coefficients," Mathematische Annalen, vol. 261, pp. 515-534, 1982 [12]). Using matrices

and T form a modified channel matrix

.

принимаемого сигнала, а на второй его вход с первого выхода узла 14 преобразования канальной 5 матрицы поступает модифицированная канальная матрица G, формируют вектор

оценок модифицированных символов модуляции. При этом используют, например, формулу (12).In the node 15 assessment of the modified modulation symbols, the first input of which from the output of the node 13 of the formation of the vector of the received signal receives the vector

the received signal, and at its second input from the first output of the channel 5 matrix conversion unit 14, a modified channel matrix G arrives, form a vector

estimates of modified modulation symbols. In this case, for example, formula (12) is used.

оценок модифицированных символов модуляции с выхода узла 15 оценки модифицированных символов модуляции поступает на первый вход узла 16 квантования, в котором выполняют квантование вектора

оценок модифицированных символов модуляции. При этом по матрице Т редукции базиса решетки и символам выбранного вида модуляции определяют модифицированные символы модуляции. Для каждого элемента вектора

определяют квантованное значение как модифицированный символ модуляции, обеспечивающий наименьшую ошибку квантования, и формируют, таким образом, вектор u, элементами которого являются квантованные оценки модифицированных символов модуляции. Ошибку квантования определяют, например, как абсолютную величину разности между квантованным и не квантованным значением оценки.Formed Vector

estimates of the modified modulation symbols from the output of the node 15 estimates of the modified modulation symbols are fed to the first input of the quantization node 16, in which the vector is quantized

estimates of modified modulation symbols. At the same time, modified modulation symbols are determined from the matrix T of the reduction of the lattice basis and the symbols of the selected type of modulation. For each element of the vector

determining the quantized value as a modified modulation symbol providing the smallest quantization error, and thus forming the vector u , whose elements are quantized estimates of the modified modulation symbols. The quantization error is defined, for example, as the absolute value of the difference between the quantized and non-quantized estimation value.

оценок символов модуляции, для чего вектор u преобразуют в вектор r действительных оценок модифицированных символов модуляции по формуле r=Tu. А затем, используя вектор r, формируют вектор

оценок символов модуляции, соответствующих N передаваемым информационным сигналам, по формуле

где r1, r2 - N-мерные векторы, составленные из первой и второй половины элементов вектора r соответственно,

- мнимая единица.The generated vector u from the output of the quantization node 16 is fed to the second input of the modulation symbol estimation node 17, in which the vector

estimates of modulation symbols, for which the vector u is converted into the vector r of real estimates of the modified modulation symbols by the formula r = Tu . And then, using the vector r , form a vector

estimates of modulation symbols corresponding to N transmitted information signals according to the formula

where r 1, r 2 - N-dimensional vectors composed of the first and second half of the elements of the vector r, respectively,

- imaginary unit.

оценок символов модуляции с выхода узла 17 оценки символов модуляции поступает на вход узла 18 оценки двоичных символов. В узле 18 оценки двоичных символов формируется последовательность двоичных символов, соответствующих вектору

. Полученная последовательность двоичных символов поступает на выход 11 блока демодуляции, а с выхода блока 11 демодуляции на вход блока 12 принятия решения. В блоке 12 принимают решение, например, путем декодирования последовательности двоичных символов. Выходной сигнал с блока 12 принятия решения поступает на выход устройства.Vector

modulation symbol estimates from the output of the modulation symbol estimation node 17 is input to the binary symbol estimation node 18. In the node 18 evaluation of binary characters is formed a sequence of binary characters corresponding to the vector

. The resulting sequence of binary characters goes to the output 11 of the demodulation block, and from the output of the demodulation block 11 to the input of the decision block 12. At block 12, a decision is made, for example, by decoding a sequence of binary symbols. The output signal from block 12 decision is fed to the output of the device.

The described prototype method and device for its implementation use the linear method of signal detection in conjunction with the reduction of the lattice basis of the channel matrix. The technique of reducing the lattice basis allows one to reduce the conditionality of the channel matrix and, as a result, to obtain improved noise immunity characteristics relative to the initial suboptimal linear method. However, this improvement is observed in the range of relatively high signal to noise ratio (SNR). At the same time, the prototype method and the device for its implementation are still significantly inferior in noise immunity to the maximum likelihood method, and in the region of low SNR values they also lose to the original linear method.

This fact is explained by the fact that an important operation of this approach is quantization, the purpose of which is to filter the estimation noise before performing the transformation, the inverse reduction of the lattice basis. However, quantization, being a nonlinear operation, is effective only when the noise deviations are small. In the region of high relative noise level, quantization leads to detection errors and reception characteristics deteriorate.

Another significant disadvantage of the prototype method and device for its implementation is low efficiency when using soft decoding. Soft decoding involves obtaining information about both the binary symbol values of the information message and the reliability of the evaluation of these values. Quantization leads to the formation of hard - quantized estimates, from which it is difficult to extract information about the reliability of the estimate. As a result, with soft decoding, the loss of the method and prototype device relative to the optimal method increases.

The objective of the invention is to increase the noise immunity of receiving a multicomponent signal, both in a radio communication system without coding, and in a radio communication system with coding for soft and hard decoding.

The problem is solved by the claimed method of receiving a multicomponent signal in a radio communication system with N transmission channels and M reception channels (options) and a device for its implementation (options), the algorithm of which is developed with practical implementation complexity and high noise immunity of reception, as in communication systems without coding , and in communication systems with coding for soft and hard decoding.

The problem is solved by the development of the proposed method for receiving a multi-component signal in a radio communication system with N transmission channels and M reception channels according to the first embodiment, in which

each of the M components of the multicomponent signal received on the corresponding receive channel is an additive mixture of N information signals, each of which is transmitted on the corresponding transmission channel, and interference,

wherein each of the N information signals corresponds to L binary symbols of the transmitted information message, takes one of m = 2 ^L values corresponding to m symbols of the selected type of modulation, and is included in each of the M components of a multicomponent signal with a weight coefficient reflecting the transmission coefficient of the spatial channel, formed by the transmission channel of this information signal and the reception channel of this component of a multicomponent signal, and it is believed that the transmission coefficients of the spatial channel in and the dispersion values of the interference M of the reception channels are known or pre-estimated, namely, that

принимаемого многокомпонентного сигнала;form a vector

a received multicomponent signal;

используя коэффициенты передачи пространственных каналов, таким образом, что между вектором

принимаемого многокомпонентного сигнала и канальной матрицей выполняется соотношениеform a channel matrix

using the transmission coefficients of the spatial channels, so that between the vector

the received multicomponent signal and the channel matrix, the relation

- вектор символов модуляции, соответствующих N информационным сигналам,Where

is a vector of modulation symbols corresponding to N information signals,

- вектор помех М каналов приема;

- interference vector M receiving channels;

в матрицу G, характеризующуюся заведомо низким числом обусловленности, для чегоtransform the channel matrix

into the matrix G , characterized by a knowingly low condition number, for which

формируют целочисленную матрицу Т с определителем, равным ±1, и формируют модифицированную канальную матрицу G, как произведение

by reducing the matrix lattice basis

form an integer matrix T with a determinant equal to ± 1, and form a modified channel matrix G , as the product

и оценивают его, используя канальную матрицу G и вектор

принимаемого сигнала, формируя таким образом вектор

оценок модифицированных символов модуляции,define the vector z of modified modulation symbols as

and evaluate it using the channel matrix G and the vector

received signal, thus forming a vector

estimates of modified modulation symbols,

оценок модифицированных символов модуляции и матрицу Т, формируют К кандидатских векторов, каждый из которых представляет собой совокупность оценок N символов модуляции, соответствующих N переданным информационным сигналам, где К≤m^N,using vector

estimates of the modified modulation symbols and the matrix T , form K candidate vectors, each of which is a set of estimates of N modulation symbols corresponding to N transmitted information signals, where K≤m ^N ,

for what

оценок модифицированных символов модуляции, определяя для каждого элемента этого вектора квантованные значения, как модифицированные символы модуляции, формируя из этих квантованных значений векторы u, каждый из которых представляет собой совокупность квантованных оценок модифицированных символов модуляции, и определяют К векторов u с наименьшими значениями погрешности квантования,quantize the vector

estimates of modified modulation symbols, defining quantized values for each element of this vector as modified modulation symbols, forming vectors u from these quantized values, each of which is a set of quantized estimates of modified modulation symbols, and determine K vectors u with the smallest quantization error values,

each of the K vectors u is multiplied on the left by the matrix T , forming the vector r

r = Tu

the elements of the obtained vectors r that do not belong to the set of all possible values of the quadrature components of the symbols of the selected type of modulation are replaced by the closest value from the given set,

thus forming K real candidate vectors r ;

each of the generated K valid candidate vectors is converted into a candidate vector according to the formula

v = r 1 + j · r 2,

- мнимая единица.where r 1, r 2 are N-dimensional vectors composed of the first and second half of the elements of the candidate vector r, respectively,

- imaginary unit.

using K formed candidate vectors, the sequence of binary symbols is evaluated, and the decision on the transmitted information message is made based on the results of the evaluation.

The problem is also solved by the claimed method of receiving a multi-component signal in a radio communication system with N transmission channels and M reception channels according to the second embodiment, in which

each of the M components of the multicomponent signal received on the corresponding receive channel is an additive mixture of N information signals, each of which is transmitted on the corresponding transmission channel, and interference,

wherein each of the N information signals corresponds to L binary symbols of the transmitted information message, takes one of m = 2 ^L values corresponding to m symbols of the selected type of modulation, and is included in each of the M components of a multicomponent signal with a weight coefficient reflecting the transmission coefficient of the spatial channel, formed by the transmission channel of this information signal and the reception channel of this component of a multicomponent signal, and it is believed that the transmission coefficients of the spatial channel in and the dispersion values of the interference M of the reception channels are known or pre-estimated, namely, that

принимаемого многокомпонентного сигнала;form a vector

a received multicomponent signal;

используя коэффициенты передачи пространственных каналов, таким образом, что между вектором

принимаемого многокомпонентного сигнала и канальной матрицей выполняется соотношениеform a channel matrix

using the transmission coefficients of the spatial channels, so that between the vector

the received multicomponent signal and the channel matrix, the relation

- вектор символов модуляции, соответствующих N информационным сигналам,

- вектор помех М каналов приема;Where

is a vector of modulation symbols corresponding to N information signals,

- interference vector M receiving channels;

, в матрицу G, характеризующуюся заведомо низким числом обусловленности, для чегоtransform the channel matrix

into the matrix G , characterized by a knowingly low condition number, for which

формируют целочисленную матрицу Т с определителем, равным ±1, и формируют модифицированную канальную матрицу G, как произведение

by reducing the matrix lattice basis

form an integer matrix T with a determinant equal to ± 1, and form a modified channel matrix G , as the product

, и оценивают его, используя канальную матрицу G и вектор

принимаемого сигнала, формируя, таким образом, вектор

оценок модифицированных символов модуляции;define the vector z of modified modulation symbols as

, and evaluate it using the channel matrix G and the vector

received signal, thus forming a vector

estimates of modified modulation symbols;

оценок модифицированных символов модуляции и матрицу Т, формируют К кандидатских векторов, каждый из которых представляет собой совокупность оценок N символов модуляции, соответствующих N переданным информационным сигналам, где K≤m^N, для чегоusing vector

estimates of the modified modulation symbols and the matrix T , form K candidate vectors, each of which is a set of estimates of N modulation symbols corresponding to N transmitted information signals, where K≤m ^N , for which

determine the set of all possible values of the quadrature components of the symbols of the selected type of modulation and for this set determine the scaling coefficient α and the shift value β so that as a result of the linear transformation a '= αa + β of each element a of this set, the set of converted values a' represents a sequence of integers , in which each subsequent integer is one more than the previous one;

оценок модифицированных символов модуляции по формулеcalculate the shift vector for the vector

estimates of modified modulation symbols according to the formula

b = T ^-1 β l

where l is a 2N-dimensional unit vector;

оценок модифицированных символов модуляции по формулеusing the scaling factor α and the shift vector b , linear transformation of the vector is performed

estimates of modified modulation symbols according to the formula

getting the vector y ;

rounding off each of the elements of the obtained vector y to the nearest integer, thus determining the vector of the first quantized values

y _Q1 = round (y),

and determining a vector of corresponding quantization error values

d ₁ = y _Q1 - y ;

determine the second vector of quantized values,

y _Q2 = y _Q1 -sign ( d ₁ ),

and also the second vector of the corresponding quantization error values, according to the formula

d ₂ = y _Q2 - y

;Where

;

Using the vectors of quantized values from _Q1, in _Q2 and vectors d _1, d ₂ corresponding to the quantization errors, form different vectors u, each of which represents a set of quantized estimates modified modulation symbols, and determine the K vectors u, having the smallest value of the total quantization errors;

each of the K vectors u is transformed by the formula

the elements of the obtained vectors r that do not belong to the set of all possible values of the quadrature components of the symbols of the selected type of modulation are replaced by the closest value from the given set

thus forming K real candidate vectors r ;

each of the generated K valid candidate vectors r is converted into a candidate vector by the formula

v = r 1 + j · r 2,

- мнимая единица,where r 1, r 2 are N-dimensional vectors composed of the first and second half of the elements of the vector r ,

- imaginary unit

using K formed candidate vectors, the sequence of binary symbols is evaluated, and the decision on the transmitted information message is made based on the results of the evaluation.

The following are examples of the sequence of actions of the proposed method according to the first and / or second implementation options, i.e. those examples of execution that are common to both variants of the method.

принимаемого сигнала формируют, принимая каждую из М компонент многокомпонентного сигнала, определяя для каждой из них комплексное число, модуль которого отражает амплитуду, а аргумент - фазу данной компоненты, формируя из данных комплексных чисел М-мерный комплексный вектор х принимаемого сигнала и преобразуя данный вектор к 2М-мерному вектору принимаемого сигнала по формулеFor example, for the method according to the first and second embodiments, the vector

the received signal is formed by taking each of the M components of the multicomponent signal, determining for each of them a complex number whose module reflects the amplitude, and the argument is the phase of this component, forming the M-dimensional complex vector x of the received signal from these complex numbers and transforming this vector to 2M-dimensional vector of the received signal according to the formula

where Re x , Im x are vectors formed from the real and imaginary parts of the elements of the vector x, respectively.

формируют таким образом, что из коэффициентов h_i,j передачи пространственных каналов, каждый из которых представляет собой комплексное число, с модулем и аргументом, отражающими изменение, соответственно, амплитуды и фазы j-го информационного сигнала, принимаемого в i-м канале приема, при его распространении по соответствующему пространственному каналу, формируют комплексную канальную матрицу Н, которую преобразуют в канальную матрицу

по формулеChannel matrix

form in such a way that from the coefficients h _{i, j the} transmission of spatial channels, each of which is a complex number, with a module and an argument reflecting the change, respectively, in the amplitude and phase of the j-th information signal received in the i-th reception channel, when it propagates through the corresponding spatial channel, a complex channel matrix H is formed , which is converted into a channel matrix

according to the formula

where Re H and Im H are matrices formed respectively from the real and imaginary parts of the elements of the matrix H.

в матрицу G перед редукцией базиса решетки матрицу

расширяют, дополняя снизу матрицей σI, где σ² - дисперсия помехи каждого из М каналов приема, I - единичная диагональная матрица размерности 2N×2N.When transforming the channel matrix

to the matrix G before the reduction of the basis of the lattice matrix

expand, supplementing the bottom with the matrix σ I , where σ ² is the variance of the interference of each of the M reception channels, I is the unit diagonal matrix of dimension 2N × 2N.

оценок модифицированных символов модуляции формируют по формулеVector

estimates of the modified modulation symbols are formed by the formula

where (.) ^-1 is the matrix inversion symbol, (.) 'is the transpose symbol.

оценок модифицированных символов модуляции формируют по формулеVector

estimates of the modified modulation symbols are formed by the formula

where (.) ^-1 is the matrix inversion symbol, (.) 'is the transpose symbol,

- вектор принимаемого сигнала, расширенный снизу 2М-мерным вектором нулей 0.

is the vector of the received signal, extended from below by a 2M-dimensional vector of zeros 0 .

оценок модифицированных символов модуляции определяют как эвклидово расстояние между векторами квантованных и неквантованных оценок.During the formation of K candidate vectors, the quantization error of the vector

estimates of modified modulation symbols are defined as the Euclidean distance between the vectors of quantized and non-quantized estimates.

оценок N символов модуляции, соответствующих N информационным сигналам, как вектор v из К кандидатских векторов, для которого решающая функция

минимальна, то естьThe sequence of binary symbols of the information message is evaluated in such a way that a vector is formed

estimates of N modulation symbols corresponding to N information signals, as a vector v from K candidate vectors, for which the decisive function

minimal, i.e.

- норма вектора,Where

is the norm of the vector,

оценок N символов модуляции, соответствующих N информационным сигналам, в соответствии с заданным видом модуляции, формируя таким образом оценки двоичных символов передаваемого информационного сообщения.perform vector demodulation

estimates of N modulation symbols corresponding to N information signals in accordance with a predetermined type of modulation, thereby forming binary symbol estimates of the transmitted information message.

The binary symbol sequence of the transmitted informational message is evaluated in such a way that

form soft estimates of the binary symbols of the information message, moreover, a soft estimate of B _{n, i of the} i-th binary symbol transmitted using the nth information signal is formed as:

,

- взаимоисключающие подмножества множества символов модуляции, которым соответствует i-й двоичный символ, равный +1 и -1 соответственно, σ² - дисперсия помехи каждой из М компонент многокомпонентного сигнала, h _n - n-й столбец комплексной канальной матрицы Н;where x is the complex vector of the received signal, s is the modulation symbol, ν _{k, j} is the jth element of the kth candidate vector v _k ,

,

- mutually exclusive subsets of the set of modulation symbols that correspond to the i-th binary symbol equal to +1 and -1, respectively, σ ² is the interference variance of each of the M components of the multicomponent signal, h _n is the nth column of the complex channel matrix H ;

form strict estimates of the binary symbols of the information message, moreover, a hard estimate b of each binary symbol is formed by a soft estimate of a given symbol in the formula

the generated soft and hard estimates of the binary symbols of the information message are used as the result of the evaluation.

The decision about the transmitted information message is made by deinterleaving and decoding the binary symbol stream using the obtained hard and soft binary symbol estimates, thus restoring the binary symbol stream of the transmitted message.

The technical problem is also solved by the claimed device for receiving a multi-component signal in a radio communication system with N transmission channels and M receiving channels according to the first embodiment, containing M antennas and, accordingly, M receiving channels for receiving a multi-component signal, a channel estimator, a demodulation unit and a decision block, the inputs of the M antennas are the signal inputs of the device, and the outputs are connected respectively to the inputs of the M receiving channels, the outputs of which are connected via a data bus to the input of the channel estimator and the first input the demodulation unit, the inputs of which are signal inputs, the second and third inputs of the demodulation unit are connected respectively to the first and second outputs of the channel estimator, which generates at the first output an estimate signal of the complex transmission coefficients of the multicomponent signal during propagation from each of the transmitting antennas to each of the receiving antennas, and at the second output, a signal for estimating the noise variance of the reception channels, the first output of the demodulation unit, which forms a hard estimate of the binary symbols of information onnogo messages, connected to the first input of the decision, the output of which is an output device,

оценок модифицированных символов модуляции и формирующий на первом выходе первый вектор квантованных оценок модифицированных символов модуляции, и узел оценки двоичных символов, при этом вход узла формирования вектора принимаемого сигнала является первым входом блока демодуляции, первый и второй входы узла преобразования канальной матрицы являются соответственно вторым и третьим входами блока демодуляции, выход узла формирования вектора принимаемого сигнала соединен с первым входом узла оценки модифицированных символов модуляции, второй вход которого соединен с первым выходом узла преобразования канальной матрицы, выход узла оценки модифицированных символов модуляции соединен с первым входом узла квантования, второй вход которого соединен со вторым выходом узла преобразования канальной матрицы, первый выход узла оценки двоичных символов является первым выходом блока демодуляции,the demodulation block contains a node for generating a vector of the received signal, a node for converting the channel matrix, a node for evaluating the modified modulation symbols, a quantization node quantizing the elements of the vector

estimates of the modified modulation symbols and generating at the first output a first vector of quantized estimates of the modified modulation symbols and a binary symbol estimation node, wherein the input of the received signal vector generation section is the first input of the demodulation block, the first and second inputs of the channel matrix transform node are second and third, respectively the inputs of the demodulation unit, the output of the received signal vector forming unit is connected to the first input of the modified modulation symbol evaluation unit the second input of which is connected to the first output of the channel matrix transform node, the output of the modified modulation symbol estimation node is connected to the first input of the quantization node, the second input of which is connected to the second output of the channel matrix transform node, the first output of the binary symbol estimation node is the first output of the demodulation block,

according to the invention

the quantization node contains a matrix inversion element, a memory element, a multiplier, a generator of valid quantized values and a quantization element, the first input of the quantization element being the first input of the quantization node, the input of the matrix inversion element is the second input of the quantization node, the output of the matrix inversion element is connected to the first input a multiplier, the second input of which is connected to the output of the memory element, the output of the multiplier is connected to the input of the generator of permissible quantized values, the outputs of which are connected with the second inputs of the quantization element, the first, second, third and fourth outputs of which form the first, second, third and fourth outputs of the quantization node, respectively, which forms the second vector of quantized estimates of the modified modulation symbols at the second output, the first and second vectors having the smallest error values quantization, and at the third and fourth outputs, respectively, forming two vectors of values of quantization errors, the second, third and fourth outputs of the quantization element form respectively second, third and fourth quantizing unit outputs,

a candidate vector forming unit is introduced into the demodulation unit, the first input of which is connected to the second output of the channel matrix conversion unit, the second, third, fourth and fifth inputs of the candidate vector forming unit are connected to the first, second, third and fourth outputs of the quantization node, the output of the forming unit candidate vectors, forming at the output of K candidate vectors, each of which is a set of estimates of N modulation symbols corresponding to N transmitted information signal, connected to the first input of the binary character evaluation unit, the second input of which is connected to the first input of the demodulation unit, the third and fourth inputs of the binary character evaluation unit are connected to the second and third inputs of the demodulation unit, the second output of the binary character evaluation unit, forming on the second the output is soft estimates of the binary symbols of the information message, each of which represents a measure of the reliability of the evaluation of the corresponding binary symbol, is the second output of the demodulation block, which is connected to the second input of the decision block. The technical problem is also solved by the inventive device for receiving a multi-component signal in a radio communication system with N transmission channels and M reception channels according to the second embodiment, containing M antennas and respectively M reception channels for receiving a multi-component signal, a channel estimation unit, a demodulation unit and a decision making unit, the inputs of the M antennas are the signal inputs of the device, and the outputs are connected respectively to the inputs of the M reception channels, the outputs of which are connected via the data bus to the input of the channel estimator and the first input the demodulation unit, the inputs of which are signal inputs, the second and third inputs of the demodulation unit are connected respectively to the first and second outputs of the channel estimator, which generates at the first output an estimate signal of the complex transmission coefficients of the multicomponent signal during propagation from each of the transmitting antennas to each of the receiving antennas, and at the second output, a signal for estimating the noise variance of the reception channels, the first output of the demodulation unit, which forms a hard estimate of the binary symbols of information at the first output onnogo messages, connected to the first input of the decision block, whose output is the output of the apparatus,

оценок модифицированных символов модуляции и формирующий на первом выходе первый вектор квантованных оценок модифицированных символов модуляции, и узел оценки двоичных символов, при этом вход узла формирования вектора принимаемого сигнала является первым входом блока демодуляции, первый и второй входы узла преобразования канальной матрицы являются соответственно вторым и третьим входами блока демодуляции, выход узла формирования вектора принимаемого сигнала соединен с первым входом узла оценки модифицированных символов модуляции, второй вход которого соединен с первым выходом узла преобразования канальной матрицы, выход узла оценки модифицированных символов модуляции соединен с первым входом узла квантования, второй вход которого соединен со вторым выходом узла преобразования канальной матрицы, первый выход узла оценки двоичных символов является первым выходом блока демодуляции,the demodulation block contains a node for generating a vector of the received signal, a node for converting the channel matrix, a node for evaluating the modified modulation symbols, a quantization node quantizing the elements of the vector

estimates of the modified modulation symbols and generating at the first output a first vector of quantized estimates of the modified modulation symbols and a binary symbol estimation node, wherein the input of the received signal vector generation section is the first input of the demodulation block, the first and second inputs of the channel matrix transform node are second and third, respectively the inputs of the demodulation unit, the output of the received signal vector forming unit is connected to the first input of the modified modulation symbol evaluation unit the second input of which is connected to the first output of the channel matrix transform node, the output of the modified modulation symbol estimation node is connected to the first input of the quantization node, the second input of which is connected to the second output of the channel matrix transform node, the first output of the binary symbol estimation node is the first output of the demodulation block,

according to the invention

the quantization unit contains a memory element, a shear vector generator, a linear transformation element and a rounding element, the first input of the shear vector generator is the second input of the quantization node, the second input of the shear vector generator is connected to the first output of the memory element, the second output of which is connected to the first input of the element linear transformation, the second input of which is connected to the output of the shear vector shaper, the third input of the linear transformation element is the first input of the quantum node the output of the linear transformation element is connected to the input of the rounding element, the first, second, third and fourth outputs of the rounding element respectively form the first, second, third and fourth outputs of the quantization node, which forms the second vector of quantized estimates of the modified modulation symbols at the second output, the first and the second vectors have the smallest quantization error values, and at the third and fourth outputs, respectively forming two quantization error vector values,

a candidate vector forming unit is introduced into the demodulation unit, the first input of which is connected to the second output of the channel matrix conversion unit, the second, third, fourth and fifth inputs of the candidate vector forming unit are connected to the first, second, third and fourth outputs of the quantization node, the output of the forming unit candidate vectors, forming at the output of K candidate vectors, each of which is a set of estimates of N modulation symbols corresponding to N transmitted information signal, connected to the first input of the binary character evaluation unit, the second input of which is connected to the first input of the demodulation unit, the third and fourth inputs of the binary character evaluation unit are connected to the second and third inputs of the demodulation unit, the second output of the binary character evaluation unit, forming on the second the output is a soft estimate of the binary symbols of the information message, each of which represents a measure of the reliability of the evaluation of the corresponding binary symbol, is the second output of the demodulation block, which is connected to the second input of the decision block.

Thus, the claimed group of inventions - the method and device (s) differs from the above method and prototype device and prior art in that instead of the traditional (known) quantization and generation of estimates of modulation symbols from the obtained quantized estimates, multi-alternative quantization and the formation of a plurality candidate vectors. The candidate vector is a set of modulation symbols that are quantized (hard) estimates of N transmitted information signals. The set of candidate vectors is composed of several of the most likely quantized estimates of each of the transmitted modulation symbols. Candidate vectors are formed in the space of modified signals, and then converted to the space of the original signals.

The resulting set of candidate vectors is used to find a solution that minimizes the decisive maximum likelihood function, or to form metrics for soft decoding.

This approach allows, firstly, to increase the noise immunity of the reception, since the choice of several vectors of quantized estimates can compensate for the detection errors associated with quantization at a high noise level.

Secondly, this makes it possible to form metrics for effective soft decoding of binary symbols of the transmitted message.

To implement the distinguishing features of the method (options), a multicomponent signal receiving device in a radio communication system with N transmission channels and M reception channels (options) is additionally introduced with a candidate vector generation unit and, accordingly, a quantization block diagram is proposed for each quantization option, which together with new compounds (bonds) ensure the implementation of the distinguishing features of the proposed method and allow you to get a better technical effect compared to known technical solutions in the art.

Thus, the claimed group of inventions created in a single inventive concept, allows to increase the noise immunity of receiving a multi-component signal, both in a radio communication system without encoding, and in a radio communication system with encoding for soft and hard decoding.

The search in the prior art did not reveal technical solutions with the claimed distinguishing features of the invention.

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

Figure 1 is a structural diagram of a MIMO system (radio communication system with N transmission channels and M channels for receiving a multi-component signal), shown for implementing the prototype method.

Figure 2 is a structural diagram of a device for receiving a multi-component signal in a MIMO system (prototype).

Figure 3 is a structural diagram of a MIMO system (radio communication system with N transmission channels and M reception channels), shown for implementing the claimed invention.

Figure 4 is a structural diagram of the inventive device for receiving a multi-component signal in a radio communication system with N transmission channels and M reception channels.

Figure 5 is a structural diagram of a quantization node for the inventive device, made according to the first embodiment.

Figure 6 is a structural diagram of a quantization node for the inventive device, made according to the second embodiment.

7 is a block diagram of a channel estimator, shown as an example implementation.

On Fig shows an example of a timing diagram of the transmission and reception of pilot signals in a MIMO system with N = 4 transmitting and M receiving antennas.

Figure 9 shows the comparative noise immunity of the proposed method of receiving a multi-component signal in a radio communication system with N transmission channels and M reception channels (MIMO), a prototype method, a maximum likelihood algorithm (MLA) and a minimum mean square error (MMSE) algorithm, obtained by computer modeling.

оценок модифицированных символов модуляции и формирующий на первом выходе первый вектор квантованных оценок модифицированных символов модуляции, и узел 18 оценки двоичных символов, при этом вход узла формирования вектора принимаемого сигнала является первым входом блока 11 демодуляции, первый и второй входы узла 14 преобразования канальной матрицы являются соответственно вторым и третьим входами блока 11 демодуляции, выход узла 13 формирования вектора принимаемого сигнала соединен с первым входом узла 15 оценки модифицированных символов модуляции, второй вход которого соединен с первым выходом узла 14 преобразования канальной матрицы, выход узла 15 оценки модифицированных символов модуляции соединен с первым входом узла квантования 16, второй вход узла 16 квантования соединен со вторым выходом узла 14 преобразования канальной матрицы, первый выход узла 18 оценки двоичных символов является первым выходом блока 11 демодуляции, согласно изобретению, узел 16 квантования содержит элемент 20 инверсии матрицы, элемент 21 памяти, умножитель 22, формирователь 23 допустимых квантованных значений и элемент 24 квантования, при этом первый вход элемента 24 квантования является первым входом узла 16 квантования, вход элемента инверсии матрицы является вторым входом узла 16 квантования, выход элемента 20 инверсии матрицы соединен с первым входом умножителя 22, второй вход которого соединен с выходом элемента 21 памяти, выход умножителя 22 соединен со входом формирователя 23 допустимых квантованных значений, выходы которого соединены со вторыми входами элемента 24 квантования, первый, второй, третий и четвертый выходы которого являются соответственно первым, вторым, третьим и четвертым выходами узла квантования, формирующего на втором выходе второй вектор квантованных оценок модифицированных символов модуляции, причем первый и второй векторы имеют наименьшие значения погрешности квантования, а на третьем и четвертом выходах формирующий соответственно два вектора значений погрешностей квантования, второй, третий и четвертый выходы элемента 24 квантования образуют соответственно второй, третий и четвертый выходы узла 16 квантования, в блок 11 демодуляции введен узел 19 формирования кандидатских векторов, первый вход которого соединен со вторым выходом узла 14 преобразования канальной матрицы, второй, третий, четвертый и пятый входы узла формирования кандидатских векторов соединены соответственно с первым, вторым, третьим и четвертым выходами узла квантования, выход узла 19 формирования кандидатских векторов, формирующего на выходе К кандидатских векторов, каждый из которых представляет собой совокупность оценок N символов модуляции, соответствующих N переданным информационным сигналам, соединен с первым входом узла 18 оценки двоичных символов, второй вход которого присоединен к первому входу блока демодуляции, третий и четвертый входы узла 18 оценки двоичных символов присоединены соответственно ко второму и третьему входам блока 11 демодуляции, второй выход узла 18 оценки двоичных символов, формирующего на втором выходе мягкие оценки двоичных символов, каждая из которых представляет меру достоверности оценки соответствующего двоичного символа, является вторым выходом блока 11 демодуляции, который соединен со вторым входом блока 12 принятия решения.A multi-component signal receiving device in a radio communication system with N transmission channels and M reception channels according to the first embodiment (Figs. 4 and 5) comprises M antennas 8-1 to 8-M and, accordingly, M reception channels 9-1 to 9-M for reception a multicomponent signal, a channel estimator 10, a demodulation unit 11, and a decision unit 12, wherein the inputs of the M antennas 8-1 to 8-M are the signal inputs of the device, and the outputs are connected respectively to the inputs of the M channels 9-1 to 9-M of the reception the outputs of which are connected via a data bus to the inputs of the channel estimation unit 10 and the first the inputs of the demodulation unit 11, the inputs of which are signal inputs, the second and third inputs of the demodulation unit are connected respectively to the first and second outputs of the channel estimator 10, which generates at the first output an evaluation signal of the complex signal transmission coefficients during propagation from each of the transmitting to each of the receiving antennas, and at the second output, a signal for estimating the noise variance of the reception channels, the first output of the demodulation unit 11, forming hard estimates of binary symbols at the first output, is connected to the first the input of the decision block 12, the output of which is the output of the device, the demodulation block 11 contains a node 13 for generating a vector of the received signal, a node 14 for converting the channel matrix, a node 15 for evaluating the modified modulation symbols, a quantization node 16 that quantizes the vector elements

estimates of the modified modulation symbols and generating at the first output a first vector of quantized estimates of the modified modulation symbols and a binary symbol estimation section 18, the input of the received signal vector forming section is the first input of the demodulation block 11, the first and second inputs of the channel matrix transform section 14 are respectively the second and third inputs of the block 11 demodulation, the output node 13 of the formation of the vector of the received signal is connected to the first input of the node 15 assessment of the modified symbols modulation, the second input of which is connected to the first output of the channel matrix transform node 14, the output of the modified modulation symbol estimator 15 is connected to the first input of the quantization node 16, the second input of the quantization node 16 is connected to the second output of the channel matrix transform node 14, the first output of the estimation node 18 binary symbols is the first output of the demodulation block 11, according to the invention, the quantization node 16 contains a matrix inversion element 20, a memory element 21, a multiplier 22, a generator 23 of valid quantized values values and quantization element 24, wherein the first input of quantization element 24 is the first input of quantization node 16, the input of the matrix inversion element is the second input of quantization node 16, the output of matrix inversion element 20 is connected to the first input of multiplier 22, the second input of which is connected to the output of the element 21 of the memory, the output of the multiplier 22 is connected to the input of the shaper 23 of the allowable quantized values, the outputs of which are connected to the second inputs of the quantization element 24, the first, second, third and fourth outputs of which are Accordingly, the first, second, third and fourth outputs of the quantization node, which forms the second vector of quantized estimates of the modified modulation symbols on the second output, the first and second vectors have the smallest quantization error values, and at the third and fourth outputs, respectively, generating two quantization error vectors, the second , the third and fourth outputs of the quantization element 24 form, respectively, the second, third and fourth outputs of the quantization unit 16, a node 19 f is introduced into the demodulation unit 11 candidate vector vectors, the first input of which is connected to the second output of the channel matrix transformation unit 14, the second, third, fourth and fifth inputs of the candidate vector generation unit are connected respectively to the first, second, third and fourth outputs of the quantization node, the output of the candidate vector generation unit 19, generating candidate vectors at the output K, each of which is a set of estimates of N modulation symbols corresponding to N transmitted information signals, connected to the first the input of the binary symbol estimation node 18, the second input of which is connected to the first input of the demodulation block, the third and fourth inputs of the binary symbol estimation node 18 are connected to the second and third inputs of the demodulation block 11, the second output of the binary symbol estimation node 18, forming the second output soft estimates of binary symbols, each of which represents a measure of the reliability of the evaluation of the corresponding binary symbol, is the second output of the demodulation block 11, which is connected to the second input of the block 12 tiya solutions.

оценок модифицированных символов модуляции и формирующий на первом выходе первый вектор квантованных оценок модифицированных символов модуляции, и узел 18 оценки двоичных символов, при этом вход узла формирования вектора принимаемого сигнала является первым входом блока 11 демодуляции, первый и второй входы узла 14 преобразования канальной матрицы являются соответственно вторым и третьим входами блока 11 демодуляции, выход узла 13 формирования вектора принимаемого сигнала соединен с первым входом узла 15 оценки модифицированных символов модуляции, второй вход которого соединен с первым выходом узла 14 преобразования канальной матрицы, выход узла 15 оценки модифицированных символов модуляции соединен с первым входом узла 16 квантования, второй вход которого является четвертым входом блока демодуляции, третий вход узла квантования соединен со вторым выходом узла 14 преобразования канальной матрицы, первый выход узла 18 оценки двоичных символов является первым выходом блока 11 демодуляции, согласно изобретению, узел 25 квантования содержит элемент памяти, формирователь 26 вектора сдвига, элемент 27 линейного преобразования и элемент 28 округления, при этом первый вход формирователя 26 вектора сдвига является вторым входом узла 16 квантования, второй вход формирователя 26 вектора сдвига соединен с первым выходом элемента 25 памяти, второй выход которого соединен с первым входом элемента 27 линейного преобразования, второй вход которого соединен с выходом формирователя 26 вектора сдвига, третий вход элемента 27 линейного преобразования является первым входом узла 16 квантования, выход элемента 27 линейного преобразования соединен с входом элемента 28 округления, первый, второй, третий и четвертый выходы элемента 28 округления образуют соответственно первый, второй, третий и четвертый выходы узла квантования, формирующего на втором выходе второй вектор квантованных оценок модифицированных символов модуляции, причем первый и второй векторы имеют наименьшие значения погрешности квантования, а на третьем и четвертом выходах формирующий соответственно, два вектора с наименьшими значениями соответствующих погрешностей квантования, в блок 11 демодуляции введен узел 19 формирования кандидатских векторов, первый вход которого соединен со вторым выходом узла 14 преобразования канальной матрицы, второй, третий, четвертый и пятый входы узла формирования кандидатских векторов соединены соответственно с первым, вторым, третьим и четвертым выходами узла квантования, выход узла 19 формирования кандидатских векторов, формирующего на выходе К кандидатских векторов, каждый из которых представляет собой совокупность оценок N символов модуляции, соответствующих N переданным информационным сигналам, соединен с первым входом узла 18 оценки двоичных символов, второй вход которого присоединен к первому входу 11 блока демодуляции, третий и четвертый входы узла 18 оценки двоичных символов присоединены соответственно ко второму и третьему входам блока 11 демодуляции, второй выход узла 18 оценки двоичных символов, формирующего на втором выходе мягкие оценки двоичных символов информационного сообщения, каждая из которых представляет меру достоверности оценки соответствующего двоичного символа, является вторым выходом блока демодуляции, который соединен со вторым входом блока 12 принятия решения.A multi-component signal receiving device in a radio communication system with N transmission channels and M reception channels according to the second embodiment (FIGS. 4 and 6) comprises M antennas 8-1 to 8-M and, accordingly, M reception channels 9-1 to 9-M for reception a multicomponent signal, a channel estimator 10, a demodulation unit 11, and a decision unit 12, wherein the inputs of the M antennas 8-1 to 8-M are the signal inputs of the device, and the outputs are connected respectively to the inputs of the M channels 9-1 to 9-M of the reception the outputs of which are connected via a data bus to the inputs of the channel estimation unit 10 and the first the inputs of the demodulation unit 11, the inputs of which are signal inputs, the second and third inputs of the demodulation unit are connected respectively to the first and second outputs of the channel estimator 10, which generates at the first output an evaluation signal of the complex signal transmission coefficients during propagation from each of the transmitting to each of the receiving antennas, and at the second output, a signal for estimating the noise variance of the reception channels, the fourth input of the demodulation block is the second input of the device to which the synchronization signal is supplied, the output of the demodulation block 11, which forms hard estimates of the binary symbols of the information message at the first output, is connected to the first input of the decision block 12, the output of which is the output of the device, the demodulation block 11 contains a received signal vector generation section 13, a channel matrix transformation section 14, a node 15 for evaluating the modified modulation symbols, a quantization node 16 that quantizes the elements of the vector

estimates of the modified modulation symbols and generating at the first output a first vector of quantized estimates of the modified modulation symbols and a binary symbol estimation section 18, the input of the received signal vector forming section is the first input of the demodulation block 11, the first and second inputs of the channel matrix transform section 14 are respectively the second and third inputs of the block 11 demodulation, the output node 13 of the formation of the vector of the received signal is connected to the first input of the node 15 assessment of the modified symbols modulation, the second input of which is connected to the first output of the channel matrix conversion unit 14, the output of the modified modulation symbol estimation unit 15 is connected to the first input of the quantization unit 16, the second input of which is the fourth input of the demodulation unit, the third input of the quantization unit is connected to the second output of the conversion unit 14 channel matrix, the first output of the binary symbol estimation unit 18 is the first output of the demodulation unit 11, according to the invention, the quantization unit 25 contains a memory element, a vector shaper 26 a shear element, a linear transformation element 27 and a rounding element 28, wherein the first input of the shear vector generator 26 is the second input of the quantization unit 16, the second input of the shear vector generator 26 is connected to the first output of the memory element 25, the second output of which is connected to the first input of the element 27 linear transformation, the second input of which is connected to the output of the shear vector generator 26, the third input of the linear transformation element 27 is the first input of the quantization unit 16, the output of the linear transformation element 27 I am connected to the input of the rounding element 28, the first, second, third and fourth outputs of the rounding element 28 form the first, second, third and fourth outputs of the quantization node, respectively, which forms the second vector of quantized estimates of the modified modulation symbols at the second output, the first and second vectors have the smallest values of the quantization error, and at the third and fourth outputs, respectively, forming two vectors with the smallest values of the corresponding quantization errors, in block 11 demodulation input there is a candidate vector generation unit 19, the first input of which is connected to the second output of the channel matrix conversion unit 14, the second, third, fourth and fifth inputs of the candidate vector generation unit are connected to the first, second, third and fourth outputs of the quantization node, the output of the forming unit 19 candidate vectors, forming at the output K candidate vectors, each of which is a set of estimates of N modulation symbols corresponding to N transmitted information signals, inen with the first input of the binary character evaluation unit 18, the second input of which is connected to the first input 11 of the demodulation unit, the third and fourth inputs of the binary character evaluation unit 18 are connected to the second and third inputs of the demodulation unit 11, the second output of the binary character evaluation unit 18 at the second output, soft estimates of the binary symbols of the information message, each of which represents a measure of the reliability of the evaluation of the corresponding binary symbol, is the second output of the demodulation block, which connected to the second input of decision block 12.

The channel estimator 10 in FIG. 7 is an exemplary embodiment for the inventive device and contains a channel matrix estimator 29 and a dispersion estimator 30, while the inputs of the channel matrix estimator 29 and the first inputs of the dispersion estimator 30 are combined to form a signal input channel estimator 10, the output of the channel matrix estimator 29 is the first output of the estimator 10 and is connected to the second input of the variance estimator 30, the output of which is the second output of the estimator 10.

1-7, all connections shown in bold lines are tires. Thus, it is shown that these compounds are designed to transmit multicomponent (vector) signals.

Next, we consider the implementation of the claimed invention with reference to the drawings.

The inventive method can be implemented in a communication system with multidimensional signals. In particular, it can be implemented in a MIMO communication system built on the principle of spatial multiplexing (or V-BLAST) (see [1], [2] and [3]).

An example of the functional structure of the transmitting and receiving devices of such a system is presented in figure 3 (for the claimed invention).

A very urgent task of MIMO communication systems is to develop an easy-to-implement signal reception algorithm with noise immunity characteristics close to optimal.

The low efficiency of simple linear reception algorithms based on the inverse of the channel matrix is explained by the random nature of this matrix, because of which it is often poorly conditioned, and the inversion operation leads to a large calculation error. This problem can be solved by an approach based on the transformation of the channel matrix by reducing the basis of the lattice. This approach uses the following methodology.

,

,

и матрицы

с помощью преобразований:The vectors x , s , n and the matrix H , connected by the basic equation of the MIMO communication system (1), are complex-valued (have complex elements). They can be represented as real-valued vectors

,

,

and matrices

using transformations:

Moreover, relation (1) also holds for real-valued vectors and matrices, i.e.

принимают значения из дискретного множества всевозможных значений квадратурных компонент символов выбранного вида модуляции. Для многих цифровых видов модуляции эти значения являются целыми числами (или могут быть преобразованы к целым числам выбором масштабирующего коэффициента). Например, для модуляции QPSK множество возможных значений элементов реально-значного вектора

{1, -1}. Соответственно, для модуляции 16QAM это множество {-3, -1, 1, 3}.Vector elements

take values from a discrete set of all possible values of the quadrature components of the symbols of the selected type of modulation. For many digital modulations, these values are integers (or can be converted to integers by choosing a scaling factor). For example, for QPSK modulation, the set of possible values of elements of a real-valued vector

{eleven}. Accordingly, for 16QAM modulation, this is the set {-3, -1, 1, 3}.

(16) может быть представлен как целочисленная линейная комбинация векторов-столбцов канальной матрицы, то естьTherefore, in the absence of noise, the vector

(16) can be represented as an integer linear combination of column vectors of the channel matrix, i.e.

- i-й столбец канальной матрицы

.Where

- i-th column of the channel matrix

.

, который вместе с канальной матрицей

формирует вектор решетки

(17), наиболее близкий к искаженному шумом вектору принимаемого сигнала

.In matrix algebra, many of these combinations are called lattices. Moreover, the matrix is called the producing one, and the set of its column vectors is called the basis of the lattice. The task of detecting a MIMO signal is presented as the task of searching in a lattice, that is, searching for such a vector

which together with the channel matrix

forms a lattice vector

(17), closest to the noise-distorted received signal vector

.

на области принятия решения в пользу той или иной точки решетки

. При одной и той же дисперсии шума достоверность решения зависит от формы этих областей, причем максимальная достоверность достигается при областях формы равностороннего параллелепипеда. Такая форма соответствует ортогональным и одинаковым по длине базисным векторам решетки. Степень приближения базиса к данному свойству можно оценить по числу обусловленности матрицы, которое определяется отношением максимального сингулярного значения данной матрицы к минимальному (см. G.H.Golub, С.F.Van Loan, "Matrix computation," Third edition. The John Hopkins University Press, Baltimore and London, 1996, 694 pp.[13]). Чем меньше число обусловленности матрицы, тем ближе ее базис к ортогональному и, соответственно, выше качество детектирования.Detection of transmitted modulation symbols is associated with the separation of a 2M-dimensional space of vector values

in the field of decision-making in favor of a particular lattice point

. With the same noise variance, the reliability of the solution depends on the shape of these regions, and the maximum reliability is achieved with the regions of the shape of an equilateral parallelepiped. Such a shape corresponds to the basis vectors of the lattice that are orthogonal and identical in length. The degree of approximation of the basis to this property can be estimated by the condition number of the matrix, which is determined by the ratio of the maximum singular value of this matrix to the minimum (see GHGolub, C.F. Van Loan, "Matrix computation," Third edition. The John Hopkins University Press, Baltimore and London, 1996, 694 pp. [13]). The smaller the condition number of the matrix, the closer its basis to the orthogonal and, correspondingly, higher detection quality.

The reduction of the matrix lattice basis is a linear transformation, which reduces the condition number of this matrix to a knowingly low value. To preserve the lattice, the linear transformation matrix must be unimodular, that is, integer with a determinant of ± 1.

формируют унимодулярную матрицу Т редукции базиса решетки. При этом может быть использован, например, известный алгоритм Lenstra-Lenstra-Lovasz (LLL) (см. [6] и [12]).When detecting by reducing the lattice basis in the matrix

form a unimodular matrix T reduction of the lattice basis. In this case, for example, the well-known Lenstra-Lenstra-Lovasz (LLL) algorithm can be used (see [6] and [12]).

преобразуют в матрицу G с помощью выражения

После этого соотношение (16) может быть представлено, какThen the matrix

convert to matrix G using the expression

After that, relation (16) can be represented as

- вектор модифицированных символов модуляции.Where

is a vector of modified modulation symbols.

After that, the problem of detecting the MIMO signal is solved with respect to the vector z of modified modulation symbols by some suboptimal method.

When solving this problem, good results were shown by method 10 of the minimum standard error (see [6], [7], [8]).

.The minimum root-mean-square error method is implemented using formulas (3), (4). In the reduced space, that is, in the space of modified vectors z , the statistical properties of the noise components differ from the properties of the elements of the vector

.

расширяют, дополняя снизу матрицей σI, где I единичная диагональная матрица размерности 2N×2N.To take this factor into account, the implementation of the minimum mean square error method with the expansion of the channel matrix is used. For this, before the formation of the unimodular matrix T, the matrix

expand by supplementing the matrix σ I from below, where I is the identity diagonal matrix of dimension 2N × 2N.

используют, при формировании матрицы Т, а затем матрицы G, то есть

This Extended Matrix

used in the formation of the matrix T , and then the matrix G , that is

оценок модифицированных символов модуляции по формулеUsing the matrix G form a vector

estimates of modified modulation symbols according to the formula

- вектор принимаемого сигнала, расширенный снизу 2М-мерным вектором нулей 0.Where

is the vector of the received signal, extended from below by a 2M-dimensional vector of zeros 0 .

оценок модифицированных символов модуляции квантуют, определяя для каждого элемента данного вектора квантованное значение, как модифицированный символ модуляции, обеспечивающий наименьшую ошибку квантования, и формируют, таким образом, вектор u квантованных оценок модифицированных символов модуляции. Ошибку квантования определяют, например, как модуль разности квантованного и не квантованного значений.Thus formed vector

estimates of the modified modulation symbols are quantized, defining a quantized value for each element of the given vector as a modified modulation symbol that provides the smallest quantization error, and thus, a vector u of quantized estimates of the modified modulation symbols is formed. The quantization error is defined, for example, as the modulus of the difference between the quantized and non-quantized values.

оценок символов модуляции, соответствующих N передаваемым информационным сигналам по формулам (13), (14).Vector u convert to vector

estimates of modulation symbols corresponding to N transmitted information signals according to formulas (13), (14).

является наиболее общим и подходит для любых видов модуляции. Первый вариант способа данного предполагаемого изобретения использует данный вариант квантования вектора

. Однако при большом числе антенн и многопозиционных видах модуляции квантование по этому алгоритму может оказаться весьма сложным.The vector quantization algorithm presented above

is the most common and suitable for all types of modulation. The first variant of the method of this alleged invention uses this variant of the quantization of the vector

. However, with a large number of antennas and multi-position types of modulation, quantization by this algorithm can be very difficult.

.On the other hand, many types of modulation (with a uniform grid of admissible values of the quadrature components of modulation symbols) allow simplified quantization of estimates. In this case, the quantization of estimates is replaced by rounding to the nearest integer, but a preliminary linear transformation of estimates — scaling and shift — is used (see [7]). This transformation is taken into account in the subsequent inverse transformation of the quantized estimates of the modified modulation symbols. The second variant of the method of the claimed invention uses this simplified vector quantization

.

Thus, the reduction of the lattice basis of the channel matrix transfers the detection problem into the space of modified signals. These signals correspond to a more “good” channel matrix, as a result of which mutual interference between these signals is minimal. This is the main factor in increasing the efficiency of suboptimal detection methods in the space of modified signals. Elements of the obtained vector of modified estimates are quantized, that is, replaced by the closest value in magnitude from the set of all possible values of the quadrature components of modulation symbols. The purpose of quantization is to filter out noise deviations and maintain the quality of the resulting estimate in the inverse transform.

However, quantization, being a nonlinear operation, is effective only when the noise deviations are small. In the region of high relative noise level, quantization leads to detection errors and reception characteristics deteriorate. This is the reason for the insufficiently high noise immunity of the prototype method, especially in the region of low signal-to-noise ratios.

The method of the claimed invention (options) differs from the prototype method described above and the prior art in that instead of the traditional quantization and generation of modulation symbol estimates from the obtained quantized estimates, multi-alternative (multivariant) quantization and the formation of multiple candidate vectors are performed. The candidate vector is a set of modulation symbols that are quantized (hard) estimates of N transmitted information signals. The set of candidate vectors is composed of several of the most likely quantized estimates of each of the transmitted modulation symbols. Candidate vectors are formed in the space of modified signals, and then converted to the space of the original signals.

The resulting set of candidate vectors is used to find a solution that minimizes the decisive maximum likelihood function, or to form metrics for soft decoding.

This approach allows, firstly, to increase the noise immunity of the reception, since the choice of a solution from several vectors of quantized estimates allows one to compensate for the detection errors associated with quantization at a high noise level.

Secondly, this makes it possible to form metrics for effective soft decoding of binary symbols of the transmitted message.

For a better understanding of the implementation of the inventive method for receiving a multi-component signal in a radio communication system with N transmission channels and M reception channels (MIMO) (Fig. 3), we briefly consider the functional structure of a multi-component signal transmission device, preparation and transmission of a multi-component signal.

The input of the encoding unit 3, which is also the input of the multicomponent signal transmission device, receives an information message intended for transmission in binary form. Block 3 encoding performs the operation of error-correcting encoding, as well as interleaving a sequence of binary characters. The latter serves to increase the stability of communication to signal fading. The stream of encoded binary symbols thus formed is fed to the input of the demultiplexing unit 4.

In block 4 of the demultiplexing, a serial stream of binary modulation symbols is divided into N streams, which are simultaneously supplied to the corresponding outputs of this block. From each of the outputs of the demultiplexing unit 4, the stream of binary symbols is input to the corresponding 5-1 - 5-N modulation unit.

Each of the blocks 5-1 to 5-N modulation contains a memory node that stores m modulation symbols in the form of complex numbers s. The modulus of each complex number s corresponds to the amplitude, and the argument is the phase of the modulation symbol.

In each block 5-1 - 5-N modulation, the input stream of binary symbols is divided into packets of L = log2 (m) binary symbols. According to this packet, which is an L-bit binary number, the corresponding modulation symbol is read from the memory node and transmitted to the output of the modulation block. Modulation methods (QPSK, M-QAM and others) are known from various sources, for example, John G. Proakis, "Digital Communication," McGrow-Hill, Third Edition [14] |.

From the output of each of the blocks 5-1 to 5-N modulation symbols of the modulation are fed to the input of the corresponding channel 6-1 - 6-N transmission. Each channel 6-1 - 6-N transmission consists of devices forming an information radio signal of a carrier frequency with amplitude and phase corresponding to the modulation symbol received at the input.

Information signals of N channels 6-1 to 6-N transmission together form a packet of N information signals. N information signals of each packet are transmitted simultaneously through N antennas 7-1 - 7-N transmission - one signal through the antenna of the corresponding transmission channel.

The invention is carried out on a multi-component signal receiving device with N transmission channels and M reception channels according to the first embodiment of FIGS. 4, 5 and 7, and according to the second embodiment, FIGS. 4, 6 and 7.

A signal is received (see FIG. 4) through the M receiving antennas 8-1 to 8-M, each of which is an input of the corresponding channel 9-1 to 9-M receiving. In each of the M channels 9-1 - 9-M receive perform the functions of signal processing at the radio frequency, synchronization, as well as transfer to the field of video frequencies and conversion to digital form. For simplicity of presentation, synchronization signals are not shown in the block diagram of the device. However, it is understood that in the inventive device for receiving a multicomponent signal, synchronization is performed by synchronization signals, which are generated in the reception channels by pilot signals (preambles) specially transmitted for this purpose. These signals respectively arrive at the corresponding inputs of the blocks, nodes and elements of the claimed device.

At the output of each of the M channels 9-1 - 9-M reception, a received signal is formed as a complex number x, the module of which reflects the amplitude, and the argument reflects the phase of this signal.

In the process of further processing the signal, the set of received signals of the M channels 9-1 - 9-M reception is considered as an M-dimensional complex vector of the received multicomponent signal x = [x ₁ , ..., x _M ] ^T.

The vector x of the received multicomponent signal thus generated is fed via the data bus to the first input of the demodulation unit 11 and to the input of the channel estimation unit 10.

The second and third inputs of the multicomponent signal demodulation unit 11 receive, respectively, estimates of the complex signal transmission coefficients during propagation from each of the transmitting antennas to each of the receiving antennas and estimates of the noise variance of the reception channels generated in the channel estimator 10.

Consider Fig. 7, in which the block diagram of the channel estimator 10 is performed; it is given as an example of execution.

In the MIMO communication system, pilot signals or preambles are used for channel estimation, which are also used for synchronization, coherent demodulation. On the receiving side, corresponding modulation symbols are known, as well as time instants and antenna numbers from which they are transmitted.

В процессе дальнейшей обработки многокомпонентного сигнала совокупность коэффициентов передачи пространственных каналов h_i,j рассматривается, как комплексная канальная матрица Н.Thus, using the pilot signals or preambles, the channel matrix estimation unit 29 estimates the transmission coefficients h _{ij of the} signal as it propagates from the jth transmitting antenna to the ith receiving antenna

In the process of further processing of the multicomponent signal, the set of transmission coefficients of the spatial channels h _{i, j is} considered as a complex channel matrix H.

Methods for estimating a channel matrix are known from the literature (see, for example, Jiun Siew, Robert Piechocki, Andrew Nix, and Simon Armor, "A Channel Estimation Method for MIMO-OFDM Systems," Center for Communications Research, University of Bristol [15]; Y. Li, "Simplified 5 channel estimation for OFDM systems with multiple transmit antennas," IEEE Trans. Wireless Comm., Vol.1, no.1, pp.67-75, Jan. 2002 [16]; Torbjorn Ekman, " Analysis of the LS Estimation Error on a Rayleigh Fading Channel, "Signals and Systems, Uppsala University, PO Box 528, SE-751 20 Uppsala, Sweden, te@signal.uu.se [17]).

In the variance estimation unit 30, the noise variance of the reception channels is estimated by known methods for estimating the statistical parameters of random processes.

For example, in a MIMO system with N = 4 transmit and M receive antennas, a pilot transmission option is possible, represented by the timing diagram of FIG. In this case, the pilot signals P _n (k) are modulation symbols known to the receiving side transmitted from each of the N transmitting antennas at finite time intervals, where n is the antenna number and k is the number of the discrete time interval. In order to avoid mutual interference, the pilot signals of various transmit antennas are separated in time. For this case, the estimate of the noise variance for the k-th receiving antenna is formed as

where n _k is the number of the antenna with which the pilot signal is transmitted on the k-th discrete time interval,

x _m (k) is the signal received by the mth receiving antenna at the kth discrete time interval,

h _{m, n} (k) is an estimate of the transmission coefficient of the propagation channel between the nth transmitting and mth receiving antennas at the kth time interval,

K is the number of discrete intervals in the evaluation period.

Usually receiving antennas are in approximately the same conditions with respect to noise and interference. Therefore, in most cases, the noise variance values of all M reception channels are the same, and upon further consideration, we will assume

Thus, at the output of the channel matrix estimation unit 29 and, accordingly, the first output of the channel estimation unit 10, an evaluation signal of complex signal transmission coefficients is generated during propagation from each of the transmitting antennas to each of the receiving antennas.

At the output of the variance estimation unit 30 and, accordingly, the second output of the channel estimator 10, a signal for evaluating the variance of the noise of the receiving channels is generated.

в соответствии с формулой (8). То есть действительные части элементов х образуют М первых элементов, а соответственно, мнимые части элементов х образуют М последних элементов вектора

. Выходной сигнал узла 13 формирования вектора принимаемого сигнала поступает на первый вход узла 15 оценки модифицированных символов модуляции.In block 11 of the demodulation (figure 4), in the node 13 of the formation of the vector of the received signal M-dimensional vector x received from the first (signal) input of block 11 of the demodulation is converted to a 2M-dimensional real vector

in accordance with formula (8). That is, the real parts of the elements x form the M first elements, and accordingly, the imaginary parts of the elements x form the M last elements of the vector

. The output signal of the node 13 forming the vector of the received signal is supplied to the first input of the node 15 evaluating the modified modulation symbols.

The node 14 of the channel matrix transformation can be performed in 2 versions - estimating the minimum mean-square error and estimating using the zero forcing method. In node 14 transform channel channel perform the following sequence of operations.

1) The complex-valued channel matrix H is converted to a real-valued matrix H _r . by the formula (10).

2a) When estimating the minimum standard error by the minimum method, the resulting matrix H _{r is} expanded in accordance with formula (11) to obtain a channel matrix

2b) When evaluating by the method of zeroing for zero, further operations use an unexpanded matrix, i.e.

формируют матрицу Т редукции базиса решетки. При этом может быть использован, например, известный алгоритм Lenstra-Lenstra-Lovasz (LLL) [6], [12].3) According to the matrix

form the matrix T of the reduction of the basis of the lattice. In this case, for example, the well-known Lenstra-Lenstra-Lovasz (LLL) algorithm [6], [12] can be used.

и Т формируют модифицированную канальную матрицу

4) Using matrices

and T form a modified channel matrix

Thus, the channel matrix transformation unit 14 forms at the first output a modified channel matrix G , which is fed to the second input of the modified modulation symbol estimation unit 15, and at the second output, forms a matrix T , which is fed to the second input of the quantization unit 16 and the first input of the node 19 the formation of candidate vectors.

принимаемого сигнала, а на второй вход - модифицированная канальная матрица G, формируют вектор

оценок модифицированных символов модуляции.In node 15 evaluations of the modified modulation symbols, the first input of which receives the vector

the received signal, and on the second input - a modified channel matrix G , form a vector

estimates of modified modulation symbols.

In this case, use the formula (12), if estimated by the method of minimum mean square error. When evaluating by zero forcing, use the formula

оценок модифицированных символов модуляции с выхода узла 15 поступает на первый вход узла 16 квантования, на второй вход которого поступила матрица Т редукции базиса решетки.Formed Vector

estimates of the modified modulation symbols from the output of the node 15 is fed to the first input of the quantization node 16, the second input of which received the matrix T of the reduction of the lattice basis.

. В отличие от способа-прототипа используется многовариантное квантование, то есть в процессе квантования для каждого элемента вектора

выбирается не одна, а несколько квантованных значений (в представленном здесь варианте реализации, по меньшей мере, два квантованных значения). На первом выходе узел 16 квантования формирует вектор квантованных оценок модифицированных символов модуляции, на втором выходе - по меньшей мере, еще один (второй) вектор квантованных оценок модифицированных символов модуляции, а на третьем и четвертом выходах - соответственно, по меньшей мере, два вектора со значениями соответствующих погрешностей квантования, второй, третий и четвертый выходы элемента 24 квантования являются соответственно вторым, третьим и четвертым выходами узла 16 квантования.In the node 16 quantization perform quantization of the elements of the vector

. In contrast to the prototype method, multivariant quantization is used, i.e., in the quantization process for each element of the vector

not one, but several quantized values are selected (in the embodiment presented here, at least two quantized values). At the first output, the quantization unit 16 forms a vector of quantized estimates of the modified modulation symbols, at the second output, at least one more (second) vector of quantized estimates of the modified modulation symbols, and at the third and fourth outputs, respectively, at least two vectors with the values of the corresponding quantization errors, the second, third and fourth outputs of the quantization element 24 are, respectively, the second, third and fourth outputs of the quantization unit 16.

From the obtained quantized values in the node 19 of the formation of candidate vectors form K (set) of candidate vectors. In this case, candidate vectors are first formed in the space of modified modulation symbols, and then they are converted to the space of the original modulation symbols.

Since the claimed invention (options) provides two options for performing quantization, we will consider them in more detail, while we consider in combination with the operation of forming candidate vectors.

по первому варианту и формирование кандидатских векторов осуществляют следующим образом (фиг.4 и 5).Quantization of vector elements

according to the first embodiment, the formation of candidate vectors is as follows (Figs. 4 and 5).

символов модуляции. Данные векторы формируют в соответствии с выбранным видом модуляции, следующим образом.The memory element 21 (figure 5) contains many real-valued vectors

modulation characters. These vectors are formed in accordance with the selected type of modulation, as follows.

All sorts of combinations of N modulation symbols are formed from the symbols s of the selected type of modulation, as the vectors s _i (i = 1, ..., m ^N ). For example, in QPSK modulation (m = 4), the modulation symbols take values from the set {-1 + 1j, -1-1j, 1 + 1j, 1-1j}.

With N = 2 transmit antennas, a memory element contains m ^N = 16 all kinds of 2-dimensional vectors

формируя таким образом множество реально-значных векторов символов модуляции, которое обозначим, как

. Векторы данного множества записывают заранее в элемент 21 памяти. Поэтому сигнал записи векторов символов модуляции на вход элемента памяти показан на фиг.5 как внутренний сигнал узла квантования (т.е. сигнал предварительной записи). Однако возможно периодическое обновление записи векторов символов модуляции по сигналам синхронизации, которые, безусловно, в устройстве приема присутствуют. Но для концентрации внимания на изложении сути изобретения (основных соединений между блоками, узлами и элементами, непосредственно касающихся изобретения), с целью разгрузки иллюстрирующих изобретение фигур, соединения сигналов синхронизации не показаны на чертежах.Each vector s from a given set is converted to a real-valued vector by the formula

thus forming a set of real-valued modulation symbol vectors, which we denote by

. The vectors of this set are recorded in advance in the memory element 21. Therefore, the recording signal of the vectors of the modulation symbols to the input of the memory element is shown in FIG. 5 as an internal signal of the quantization node (i.e., the pre-recording signal). However, it is possible to periodically update the recording of modulation symbol vectors by synchronization signals, which, of course, are present in the receiving device. But to concentrate on the presentation of the essence of the invention (the main connections between blocks, nodes and elements directly related to the invention), for the purpose of unloading the figures illustrating the invention, the synchronization signal connections are not shown in the drawings.

From the second input of the quantization node, the signal (matrix T ) is fed to the input of the matrix inversion element 20, which performs the inversion T ^{-1 of the} matrix of the reduction of the lattice basis.

символов модуляции, которые поступают на его первый вход. В умножителе 22 каждый вектор из множества

символов модуляции умножают слева на инверсию матрицы T ^-1 редукции базиса решетки:

и таким образом формируют множество всевозможных векторов модифицированных символов модуляции (обозначенное далее как Z). Векторы множества Z с выхода умножителя 22 поступают на вход формирователя 23 квантованных значений.The output signal of the matrix inversion element 20 is supplied to the second input of the multiplier 22. According to this signal, the multiplier 22 reads vectors from the memory element

modulation symbols that arrive at its first input. In the multiplier 22, each vector from the set

modulation symbols are multiplied on the left by the inverse of the matrix T ^{-1 of the} reduction of the lattice basis:

and in this way a variety of all kinds of vectors of modified modulation symbols (indicated below as Z) are formed. The vectors of the set Z from the output of the multiplier 22 are fed to the input of the shaper 23 of the quantized values.

In the shaper 23 of the quantized values, from the first elements of all the vectors of the set Z, Y ₁ - the 1st set of valid quantized values of the modified modulation symbols is formed, including in this set one copy of each of the different values of the first elements of the vectors z ∈ Z.

Similarly form Y ₂ - the 2nd set of valid quantized values of the modified modulation symbols from the second elements of all vectors of the set Z, and so on. That is, in this way, each nth set Y _{n of} admissible quantized values of the modified modulation symbols is formed from n-elements of all vectors of the set Z with n = 1, ... 2N.

оценок модифицированных символов модуляции.The generated sets Y ₁ , ... Y _2N from the outputs of the shaper 23 of the admissible quantized values are supplied to the second inputs of the quantization element 24, the first input of which receives the vector

estimates of modified modulation symbols.

, вектора

определяют, по меньшей мере, по два квантованных значения (y_Q1)_n и (y_Q2)_n, как два значения из множества Y_n, обеспечивающие наименьшие значения абсолютной погрешности квантования, определяемой как разность между квантованным и неквантованным значением, то естьIn quantization element 24, for each nth element

, vector

at least two quantized values (y _Q1 ) _n and (y _Q2 ) _n are determined as two values from the set Y _n providing the smallest values of the absolute quantization error, defined as the difference between the quantized and non-quantized value, i.e.

where Y ' _{n is the} set Y _n after excluding from it the value (y _Q1 ) _n ,

are the values of the absolute quantization error corresponding to the quantized values (y _Q1 ) _n and (y _Q2 ) _n .

Thus, two vectors of quantized values are determined

y _Q1 = [(y _Q1 ) ₁ , ... (y _Q1 ) _2N ] ^T and y _Q2 = [(y _Q2 ) ₁ , ... (y _Q2 ) _2N ] ^T

and two corresponding quantization error vector: d ₁ = [d _1,1 , ... d _1,2N ] ^T and d ₂ = [d _2,1 , ... d _2,2N ] ^T.

These vectors of quantized values and corresponding quantization errors are the result of a quantization operation.

The vectors y _Q1, at _Q2 of quantized values and the vectors corresponding to the error quantizing d _1, d ₂ received respectively at the first, second, third and fourth quantizing element 24 yields, which are simultaneously the first, second, third and fourth quantizing unit 16 outputs.

From the output of node 16, the vectors of quantized values and the vectors of the corresponding quantization errors are received respectively at the second, third, fourth, and fifth inputs of the node 19 for forming candidate vectors.

In the node 19 of the formation of candidate vectors determine K vectors u with the smallest values of the quantization error. This is done, for example, as follows.

From the vectors in _Q1, and _Q2 quantized values form all possible vectors u, where each n-th u _n element adopts one of two values {(at _{Q1) n,} (at _{Q2) n},} where (in _Q1), ( y _Q2 ) are the nth elements of the vectors of _Q1 and y _Q2, respectively.

вектором u.For each of the vectors u thus formed, a corresponding quantization error value D ( u ) is generated. Moreover, the quantization error is determined as any monotonic function (for example, a square) from the Euclidean distance between the vector

vector u .

There are other options for determining the quantization error, for example, as the total absolute error:

From the vectors u choose K <= m ^N vectors for which the values of the quantization error D ( u ) are minimal.

Each of the K vectors u obtained is multiplied on the left by the matrix T , that is, r = Tu .

Elements of the obtained vectors r , which do not belong to the set of all possible values of the quadrature components of the symbols of the selected type of modulation, are replaced by the closest value from the given set. For many types of modulation having a uniform grid of admissible values of the quadrature components of modulation symbols, this operation can be performed as a restriction operation in accordance with the formula

where i = 1, ... 2N, a _max and a _{max is the} maximum and, accordingly, the minimum value from all possible values of a quadrature components of modulation symbols. Thus, many valid (real-valued) candidate vectors are formed.

по формулеEach of the valid candidate vectors r thus obtained is converted into a candidate vector

according to the formula

- мнимая единица.where r 1, r 2 are N-dimensional vectors composed of the first and second half of the elements of the candidate vector r, respectively,

- imaginary unit.

Thus formed, To the candidate vectors go to the output of the node forming the candidate vectors 19.

The quantization and formation of candidate vectors according to the second embodiment (Figs. 4 and 6) is carried out as follows.

The memory element 25 (Fig. 6) contains a scaling factor α and a shift amount b of modulation symbols previously recorded therein. Therefore, the write signal to the input of the memory element 25 is shown in FIG. 6 as an internal signal of a quantization unit (i.e., a pre-record signal). However, it is possible to periodically update the recording according to synchronization signals, which, of course, are present in the receiving device, but are not shown in the drawings for simplicity.

The values of α and b depend on the type of modulation and are determined by the set of possible values of the quadrature components of the modulation symbols a. The values of α and b are chosen so that, as a result of the linear transformation a '= αa + β, the set of converted values a' of the quadrature components of the modulation symbols represents a sequence of integers in which each subsequent integer is one more than the previous one.

заменить округлением элементов преобразованного вектора у до ближайшего целого числа.For example, the set of all possible values of the quadrature components and 16QAM modulation symbols is {-3, -1, 1, 3}. For α = 0.5 and b = 0.5, the linear transformation a '= αa + β leads this set to the set: {-1, 0, 1, 2}. Such a transformation allows the quantization of the values of the elements of the vector

replace by rounding the elements of the transformed vector y to the nearest integer.

с первого входа узла 16 квантования.In the shaper vector 26 of the shift according to the signals received at its first and second inputs (respectively, from the second input of the quantization node and the first output of the memory element 25), a shift vector b is generated by the formula b = β T ^-1 1 , where 1 is 2N- dimensional vector of units. The generated shear vector b from the output of the shear vector shaper 26 goes to the second input of the linear transformation element 27, the first input of which receives the scaling factor α from the second output of the memory element 25, and the values of the vector elements to the third input

from the first input of the quantization node 16.

In the element 27 of the linear transformation form the converted vector for the estimates of the modified modulation symbols according to the formula: y = α · z + b .

The transformed vector of the estimates of the modified modulation symbols from the output of the linear transformation element 27 is fed to the input of the rounding element 28.

In rounding element 28

- rounding operate each element of the resulting vector y to the nearest whole number, thereby determining the vector y of the first quantized value _Q1, and determining the vector d ₁ error values corresponding to the quantization by the formulas:

y _Q1 = round ( y ),

- determine the second vector of quantized values of _Q2 , as well as the second vector d _{2 of the} corresponding values of the quantization error, by the formulas: y _Q2 = y _Q1 -sign ( y _Q1 - y ),

.Where

.

The vectors y _Q1 , y _Q2 , d ₁ , d ₂ are the results of the quantization operation.

The vectors of the quantized values y _Q1 , y _Q2 and the vectors d ₁ , d _{2 of the} corresponding quantization errors come respectively from the first, second, third and fourth outputs of the rounding element 28 to the first, second, third and fourth outputs of the quantization unit 16.

From the first, second, third, and fourth outputs of the quantization unit 16, the vectors of quantized values y _Q1 , y _Q2 and the vectors d ₁ , d _{2 of the} corresponding quantization errors are supplied to the second, third, fourth, and fifth inputs of the candidate vector generation unit 19, respectively.

In the node 19 for the formation of candidate vectors, K vectors u are determined having the smallest values of the total quantization error. This is done using the vectors of quantized values of y _Q1 , y _Q2 and vectors d ₁ , d _{2 of the} corresponding quantization errors, similarly to the corresponding procedure described for the first variant of the method. Candidate vectors are formed from the generated K vectors u by performing the following operations.

1) Each vector u of quantized estimates of the modified modulation symbols is transformed by the formula

2) The operation of restricting the elements of each of the obtained vectors r is performed in accordance with formula (27) and, thus, the real candidate vectors r are formed .

3) Each of the actual candidate vectors r obtained in this way is transformed into the candidate vector v by the formula (28).

Thus formed, K candidate vectors are fed to the output of the candidate vector forming unit 19, the output of which is respectively sent to the first input of the binary symbol evaluation unit 18 (Fig. 4), the second input of which received a signal from the first signal input of the demodulation unit 11, the third input received a channel matrix estimation signal, and the fourth input a signal for evaluating the noise variance of the reception channels.

At the node 18, the binary symbol estimates, using the K generated candidate vectors, evaluate the binary symbol sequence of the information message.

There are two options for the operation of the node 18 evaluation of binary characters. The first option is similar to the prototype and consists in the formation of only strict estimates of binary characters. It is suitable for cases of hard decoding or complete lack of coding in the communication system. However, in this embodiment, the inventive method gives a significant gain relative to the prototype.

The second option includes the formation of soft estimates of binary characters. This option allows you to get the highest noise immunity of the connection.

оценок N символов модуляции, соответствующих N информационным сигналам, как кандидатский вектор, для которого решающая функция

минимальна, то естьIn the first embodiment, the sequence of binary symbols of the transmitted information message is evaluated in such a way that a vector is formed

estimates of N modulation symbols corresponding to N information signals, as a candidate vector for which the decisive function

minimal, i.e.

- норма вектора.Where

is the norm of the vector.

оценок N символов модуляции, соответствующих N информационным сигналам, в соответствии с заданным видом модуляции, формируя, таким образом, оценки двоичных символов передаваемого информационного сообщения.After that, vector demodulation is performed.

estimates of N modulation symbols corresponding to N information signals in accordance with a predetermined type of modulation, thereby forming binary symbol estimates of the transmitted information message.

In the second embodiment, the sequence of binary symbols of the transmitted informational message is evaluated in such a way that soft estimates of the binary symbols of the informational message are generated, and the soft estimate _{Bn, i of the} i-th binary symbol transmitted by the nth information signal is formed as:

,

- взаимоисключающие подмножества множества символов модуляции, которым соответствует i-й двоичный символ равный +1 и -1 соответственно, σ² - дисперсия помехи каждой из М компонент многокомпонентного сигнала, h _n - n-й столбец комплексной канальной матрицы Н,where x is the complex vector of the received signal, s is the modulation symbol, v _{k, j} is the jth element of the kth candidate vector v _k ,

,

- mutually exclusive subsets of the set of modulation symbols that correspond to the i-th binary symbol equal to +1 and -1, respectively, σ ² is the interference variance of each of the M components of the multicomponent signal, h _n is the nth column of the complex channel matrix H ,

form strict estimates of the binary symbols of the information message, moreover, a hard estimate b of each binary symbol is formed by a soft estimate of a given symbol in the formula

the hard and soft binary symbol estimates generated, respectively, from the first and second outputs of the binary symbol evaluation node are received respectively at the first and second outputs of the demodulation block 11.

From the first and second outputs of the demodulation block 11, hard and soft binary symbol estimates are received respectively at the first and second inputs of the decision block 12.

In decision block 12, based on the generated hard and soft binary symbol estimates, a decision is made about the transmitted information message by performing operations opposite to those performed in the transmitter coding block. For example, if encoding and interleaving of a stream of binary characters of an initial message was performed in a coding unit, then in a decision block, deinterleaving and decoding of a stream of binary characters is performed using the obtained hard and soft estimates of binary characters, thus restoring the stream of binary characters of the transmitted message. The output of decision block 12 is the output of the device.

To assess the noise immunity characteristics of the reception by the method of the claimed invention, computer simulation was performed. In this case, we evaluated the dependence of the bit error rate BER (bit error rate) on E _B / N ₀ - the ratio of the bit energy of the signal E _B to the noise power spectral density N ₀ . This dependence was evaluated for the method of the claimed invention, the prototype method, the maximum likelihood algorithm (MLA) and the minimum mean square error algorithm (MMSE).

Simulation conditions. The number of transmitting and receiving antennas N = 4, M = 4. Type of QPSK modulation. Convolutional coding with parameters K = 9, coding rate 1/2, code block size 192 bits. A communication channel with white Gaussian noise and Rayleigh fading, and the channel matrix H is a set of independent complex random variables with a Gaussian distribution of zero mean and unit dispersion.

The results are presented graphically in FIG. 9. From the graphs it can be seen that the prototype has an advantage in noise immunity relative to MMSE only in the region of high signal-to-noise ratios (Е _В / N ₀ > = 6 dB).

You can also see a significant advantage in noise immunity of the method of the claimed invention. At the level of BER = 0.001, the energy gain relative to the prototype is ≈6 dB.

This advantage is achieved due to multivariate quantization and the formation of several candidate vectors, as well as due to soft decoding performed using the generated metrics of soft solutions.

A comparison of the complexity of the proposed invention, prototype and maximum likelihood algorithm can be made by determining the asymptotic estimate of the number of computational operations necessary to receive one packet of simultaneously transmitted information signals. Using the results of complexity assessment of the lattice basis reduction procedure (see Wai No Mow, "Universal Lattice Decoding: Principle and Recent Advances," Dept. of EEE, Hong Kong University of Science and Technology, Clear Water Bay, Hong Kong SAR, CHINA, available: http://www.ee.ust.hk/~eewhmow/download [18]), we got the following estimate of the number of computational operations needed to implement the algorithms in question.

L · 2 ^LN · O (N ³ ) - for the maximum likelihood algorithm,

O (N ³ · M + N ⁴ ) + K · 2 ^L · L · O (M · N ² + M · N) - for the proposed method,

About (N ³ · M + N ⁴ ) - for the prototype,

where N, M are the number of transmitting and receiving antennas, respectively, K is the number of candidate vectors of the proposed method, L is the number of bits per modulation symbol, the expression O (n ^k m ^p ) denotes a polynomial function of n and m, and n ^k m ^p defines term of the polynomial of the highest degree.

These estimates show that the MLA algorithm has the highest - exponential with respect to L and N - complexity. The proposed algorithm has a lower complexity - polynomial with respect to the number of antennas.

The number of candidate vectors K determines the quality of demodulation of the proposed method, but, as a rule, the condition K = 2 ^LN is fulfilled, where 2 ^LN is the set of solution search vectors by the MLA method. Therefore, the complexity of the proposed algorithm is significantly lower than the complexity of the MLA. The complexity of the proposed algorithm exceeds the computational complexity of the prototype by the value of K · 2 ^L · L · O (M · N ² + M · N), which is determined by the operations of generating candidate vectors and soft decision metrics.

A rough estimate of the number of complex multiplication operations needed to demodulate a packet of simultaneously transmitted information signals, with N _T = N _R = 4, L = 4 (16-QAM modulation), K = 12, gives the following result.

5,200,000 for the maximum likelihood algorithm; 14000 for prototype 76000 for the invention.

This example shows that in a system with 4 receiving and transmitting antennas when using 16-QAM modulation, the MLA algorithm must perform a very large number of complex multiplication operations: 5200000. At the same time, for the implementation of the invention, the required number of similar operations is significantly lower than 76000, which significantly reduces implementation complexity.

Thus, the method and apparatus of the claimed invention can be used in a MIMO communication system.

This is not limited to the possibility of using the present invention. For example, for communication channels with a wide band where the problem of frequency selectivity arises, it is very promising to use MIMO technology in combination with orthogonal frequency division division (OFDM) (see IEEE P802.16-REVd / standard for radio communication systems D5-2004 Part 16: Air Interface for Fixed Broadband Wireless Access Systems, 2004 [19]). In this case, the demodulation algorithm corresponding to the presented method for receiving a multi-component signal is used in each of the orthogonal frequency subchannels. Moreover, in the receiving device, the demodulation unit 11 (Fig. 2) is implemented in the channel of each of the subcarrier frequencies.

Thus, the claimed group of inventions - the method and device (options) allows

firstly, to increase the noise immunity of the reception, since the choice of several vectors of quantized estimates makes it possible to compensate for the detection errors associated with quantization at a high noise level;

secondly, it makes it possible to form metrics for effective soft decoding of binary symbols of the transmitted message and thereby obtain the highest noise immunity. In general, with the relatively low complexity of the implementation, the algorithm allows one to realize noise immunity very close to the noise immunity of the optimal algorithm (maximum likelihood algorithm).

Therefore, the claimed group of inventions created in a single inventive concept, allows to increase the noise immunity of receiving a multi-component signal, both in a radio communication system without encoding and in a radio communication system with encoding with soft and hard decoding.

Claims (21)

1. A method of receiving a multicomponent signal in a radio communication system with N transmission channels and M reception channels, in which each of the M components of the multicomponent signal received on its corresponding reception channel is an additive mixture of N information signals, each of which is transmitted on its corresponding channel transmission, and interference, while each of the N information signals corresponds to L binary symbols of the transmitted information message, takes one of m = 2 ^L values corresponding to m symbols selected This type of modulation, and is included in each of the M components of a multicomponent signal with a weight coefficient reflecting the transmission coefficient of the spatial channel formed by the transmission channel of this information signal and the reception channel of this component of the multicomponent signal, and it is believed that the transmission coefficients of spatial channels and interference dispersion values are M the reception channels are known or pre-evaluated, namely, that form a vector

the received multicomponent signal, form a channel matrix

using the transmission coefficients of the spatial channels, so that between the vector

the received multicomponent signal and the channel matrix, the relation

Where

is a vector of modulation symbols corresponding to N information signals,

- - interference vector M reception channels,
transform the channel matrix

into a matrix G , characterized by a knowingly low condition number, for which, by reducing the matrix lattice basis

form an integer matrix T with a determinant equal to ± 1, and form a modified channel matrix G , as the product

define the vector z of modified modulation symbols as

and evaluate it using the channel matrix G and the vector

received signal, thus forming a vector

estimates of modified modulation symbols using a vector

estimates of the modified modulation symbols and the matrix T , form K candidate vectors, each of which is a set of estimates of N modulation symbols corresponding to N transmitted information signals, where K≤m ^N , for which the vector is quantized

estimates of modified modulation symbols, defining quantized values for each element of this vector as modified modulation symbols, forming vectors u from these quantized values, each of which is a set of quantized estimates of modified modulation symbols, and determine K vectors u with the smallest quantization error values, each of the K vectors u is multiplied on the left by the matrix T , forming the vector r
r = Tu
the elements of the obtained vectors r , which do not belong to the set of all possible values of the quadrature components of the symbols of the selected modulation type, are replaced by the closest value from the given set, thereby forming K real candidate vectors r , each of the received K real candidate vectors is converted into a candidate vector according to the formula
v = r 1 + j · r 2,
where r 1, r 2 - N-dimensional vectors composed of the first and second half of the elements of the candidate vector r , respectively,

- an imaginary unit, using K generated candidate vectors, evaluate the sequence of binary characters, and according to the results of the assessment, they decide on the transmitted information message. 2. The method according to claim 1, characterized in that the vector

the received signal is formed by taking each of the M components of the multicomponent signal, determining for each of them a complex number whose module reflects the amplitude, and the argument is the phase of this component, forming from the complex numbers M-dimensional complex vector x of the received signal, and transforming this vector to the 2M-dimensional vector of the received signal by the formula

where Re x , Im x are vectors formed from the real and imaginary parts of the elements of the vector x , respectively. 3. The method according to claim 1, characterized in that the channel matrix

form in such a way that from the coefficients h _{i, j the} transmission of spatial channels, each of which is a complex number, with a module and an argument reflecting the change, respectively, in the amplitude and phase of the j-th information signal received in the 1st reception channel, when it propagates through the corresponding spatial channel, a complex channel matrix H is formed , which is converted into a channel matrix

according to the formula

where Re H and Im H are matrices formed, respectively, from the real and imaginary parts of the elements of the matrix H. 4. The method according to claim 1, characterized in that when converting the channel matrix

to the matrix G before the reduction of the basis of the lattice matrix

expand, supplementing the bottom with the matrix σ I , where σ ² is the variance of the interference of each of the M reception channels, I is the unit diagonal matrix of dimension 2N × 2N. 5. The method according to claim 1, characterized in that the vector

estimates of the modified modulation symbols are formed by the formula

where (.) ^-1 is the matrix inversion symbol, (.) 'is the transpose symbol. 6. The method according to claim 1, characterized in that the vector

estimates of the modified modulation symbols are formed by the formula

where (.) ^-1 is the matrix inversion symbol, (.) 'is the transpose symbol,

is the vector of the received signal, extended from below by a 2M-dimensional vector of zeros 0 . 7. The method according to claim 1, characterized in that when forming K candidate vectors, the quantization error of the vector

estimates of modified modulation symbols determine how the Euclidean distance between the vectors of quantized and non-quantized estimates. 8. The method according to claim 1, characterized in that the sequence of binary characters of the information message is evaluated in such a way that a vector is formed

estimates of N modulation symbols corresponding to N information signals, as a vector v from K candidate vectors, for which the decisive function

minimal, i.e.

Where

is the norm of the vector,
perform vector demodulation

estimates of N modulation symbols corresponding to N information signals in accordance with a predetermined type of modulation, thereby forming binary symbol estimates of the transmitted information message. 9. The method according to claim 1, characterized in that the sequence of binary characters of the transmitted informational message is evaluated in such a way that they form soft estimates of the binary characters of the informational message, moreover, a soft estimate In _{n, i of the} i-th binary character transmitted using the nth information signal, form how

,
where x is the complex vector of the received signal, s is the modulation symbol, ν _{k, j} is the jth element of the kth candidate vector ν _k ,

,

- mutually exclusive subsets of the set of modulation symbols that correspond to the i-th binary symbol equal to +1 and -1, respectively, σ ² is the interference variance of each of the M components of the multicomponent signal, h _n is the nth column of the complex channel matrix H , form tight estimates binary symbols of the information message, moreover, a hard estimate b of each binary symbol is formed by a soft estimate B of this symbol by the formula

soft and hard binary symbol estimates formed are used as a result of the evaluation. 10. The method according to claim 1, characterized in that the decision on the transmitted information message is made by deinterleaving and decoding the binary symbol stream using the obtained hard and soft binary symbol estimates, thereby restoring the binary symbol stream of the transmitted message. 11. A method for receiving a multicomponent signal in a radio communication system with N transmission channels and M reception channels, in which each of the M components of the multicomponent signal received on its corresponding reception channel is an additive mixture of N information signals, each of which is transmitted on its corresponding channel transmission, and interference, wherein each of the N data signals corresponding to L symbols transmitted binary information message takes one of the m = 2 ^L values corresponding to the selected symbols m This type of modulation, and is included in each of the M components of a multicomponent signal with a weight coefficient reflecting the transmission coefficient of the spatial channel formed by the transmission channel of this information signal and the reception channel of this component of the multicomponent signal, and it is believed that the transmission coefficients of spatial channels and interference dispersion values are M the reception channels are known or pre-evaluated, namely, that form a vector

the received multicomponent signal, form a channel matrix

using the transmission coefficients of the spatial channels, so that between the vector

the received multicomponent signal and the channel matrix, the relation

Where

is a vector of modulation symbols corresponding to N information signals,

- interference vector M receive channels, transform the channel matrix

into the matrix G , which is characterized by a knowingly low condition number, for which, by reducing the lattice basis of the matrix

form an integer matrix T with a determinant equal to ± 1, and form a modified channel matrix G , as the product

define the vector z of modified modulation symbols as

and evaluate it using the channel matrix G and the vector

received signal, thus forming a vector

estimates of modified modulation symbols using a vector

of estimates of the modified modulation symbols and the matrix T , form K candidate vectors, each of which is a set of estimates of N modulation symbols corresponding to N transmitted information signals, where K≤m ^N , for which a set of all possible values of quadrature components of symbols of the selected type of modulation is determined and for they determine the scaling factor α and the shift value β so that, as a result of a linear transformation of each element a of this set a '= αa + β, the set of transforms Of the given values a 'was a sequence of integers in which each subsequent integer is one more than the previous one, the shift vector for the vector is calculated

estimates of modified modulation symbols according to the formula
b = T ^-1 β 1 ,
where 1 - 2N-dimensional vector of units,
using the scaling factor α and the shift vector b , linear transformation of the vector is performed

estimates of modified modulation symbols according to the formula

getting the vector y , rounding off each of the elements of the obtained vector y to the nearest integer, thus determining the vector of the first quantized values
y _Q1 = round ( y ),
and determining a vector of corresponding quantization error values
d ₁ = y _Q1 - y ,
determine the second vector of quantized values,
y _Q2 = y _Q1 -sign ( d ₁ ),
as well as a second vector of corresponding quantization error values
d ₂ = y _Q2 - y ,

using the vectors of quantized values y _q1 , y _Q2 , and vectors d ₁ , d _{2 of the} corresponding quantization errors, form different vectors u , each of which is a set of quantized estimates of the modified modulation symbols, and determine K vectors u having the smallest values of the total quantization error , each of the K vectors and transform by the formula

the elements of the obtained vectors r , which do not belong to the set of all possible values of the quadrature components of the symbols of the selected type of modulation, are replaced by the closest value from the given set, thus forming K real candidate vectors, each of the K real candidate vectors thus obtained is converted into a candidate formula vector
v = r 1 + j · r 2,
where r 1, r 2 are N-dimensional vectors composed of the first and second half of the elements of the vector r ,

- an imaginary unit, using K generated candidate vectors, evaluate the sequence of binary characters, and according to the results of the assessment, they decide on the transmitted information message. 12. The method according to claim 11, characterized in that the vector

the received signal is formed by taking each of the M components of the multicomponent signal, determining for each of them a complex number whose module reflects the amplitude, and the argument is the phase of this component, forming from the complex numbers M-dimensional complex vector x of the received signal, and transforming this vector to the 2M-dimensional vector of the received signal by the formula

where Re x , Im x are vectors formed from the real and imaginary parts of the elements of the vector x , respectively. 13. The method according to claim 11, characterized in that the channel matrix

form in such a way that from the coefficients h _{i, j the} transmission of spatial channels, each of which is a complex number, with a module and an argument reflecting the change, respectively, in the amplitude and phase of the j-th information signal received in the i-th reception channel, when it propagates through the corresponding spatial channel, a complex channel matrix H is formed , which is converted into a channel matrix

according to the formula

where Re H and Im H are matrices formed, respectively, from the real and imaginary parts of the elements of the matrix H. 14. The method according to claim 11, characterized in that when converting the channel matrix

to the matrix G before the reduction of the basis of the lattice matrix

expand, supplementing the bottom with the matrix σ I , where σ ² is the variance of the interference of each of the M reception channels, I is the unit diagonal matrix of dimension 2N × 2N. 15. The method according to claim 11, characterized in that the vector

estimates of the modified modulation symbols are formed by the formula

where (.) ^-1 is the matrix inversion symbol, (.) 'is the transpose symbol,

is the vector of the received signal, extended from below by a 2M-dimensional vector of zeros 0 . 16. The method according to claim 11, characterized in that when forming K candidate vectors, the quantization error of the vector

estimates of modified modulation symbols determine how the Euclidean distance between the vectors of quantized and non-quantized estimates. 17. The method according to claim 11, characterized in that the sequence of binary characters of the information message is evaluated in such a way that a vector is formed

estimates of N modulation symbols corresponding to N information signals, as a vector v from K candidate vectors, for which the decisive function

- minimal, i.e.

Where

is the norm of the vector,
perform vector demodulation

estimates of N modulation symbols corresponding to N information signals in accordance with a predetermined type of modulation, thereby forming binary symbol estimates of the transmitted information message. 18. The method according to claim 11, characterized in that the sequence of binary symbols of the transmitted informational message is evaluated in such a way that soft estimates of the binary symbols of the informational message are formed, moreover, the soft estimate _{Bn, i} i-ro of the binary symbol transmitted by the nth information signal, form how

where x is the complex vector of the received signal, s is the modulation symbol, ν _{k, j} is the jth element of the kth candidate vector v _k ,

,

are mutually exclusive subsets of the set of modulation symbols to which the ith binary symbol is equal to +1 and -1, respectively, σ ² is the interference variance of each of the M components of the multicomponent signal, h _n , is the nth column of the complex channel matrix H , form estimates of the binary symbols of the information message, and a hard estimate b of each binary symbol is formed by a soft estimate B of this symbol by the formula

soft and hard binary symbol estimates formed are used as a result of the evaluation. 19. The method according to claim 11, characterized in that the decision on the transmitted information message is made by deinterleaving and decoding the binary symbol stream using the obtained hard and soft binary symbol estimates, thereby restoring the binary symbol stream of the transmitted message. 20. A device for receiving a multicomponent signal in a radio communication system with N transmission channels and M reception channels, containing M antennas and, respectively, M reception channels for receiving a multicomponent signal, a channel estimator, a demodulation unit, and a decision unit, while the inputs of the M antennas are signal inputs devices, and the outputs are connected respectively to the inputs of the M receiving channels, the outputs of which are connected via a data bus to the input of the channel estimator and the first input of the demodulation unit, the inputs of which are signal inputs and, the second and third inputs of the demodulation block are connected respectively to the first and second outputs of the channel estimator, which forms at the first output an estimate signal for the complex transmission coefficients of the multicomponent signal during propagation from each of the transmitting antennas to each of the receive antennas, and at the second output, an estimate signal the noise dispersion of the reception channels, the first output of the demodulation block, which forms a hard estimate of the binary symbols of the information message at the first output, is connected to the first input of the decision block the output of which is the output of the device, the demodulation unit contains a node for generating a vector of the received signal, a node for converting the channel matrix, a node for evaluating the modified modulation symbols, a quantization node quantizing the vector elements

estimates of the modified modulation symbols and generating at the first output a first vector of quantized estimates of the modified modulation symbols and a binary symbol estimation node, wherein the input of the received signal vector generation section is the first input of the demodulation block, the first and second inputs of the channel matrix transform node are second and third, respectively the inputs of the demodulation unit, the output of the received signal vector forming unit is connected to the first input of the modified modulation symbol evaluation unit the second input of which is connected to the first output of the channel matrix transform node, the output of the modified modulation symbol estimation node is connected to the first input of the quantization node, the second input of which is connected to the second output of the channel matrix transform node, the first output of the binary symbol estimation node is the first output of the demodulation block, characterized in that the quantization unit contains an inversion element of the matrix, a memory element, a multiplier, a generator of valid quantized values and a quantization element, wherein the first input of the quantization element is the first input of the quantization node, the input of the inversion element of the matrix is the second input of the quantization node, the output of the inversion of the matrix is connected to the first input of the multiplier, the second input of which is connected to the output of the memory element, the output of the multiplier is connected to the input of the generator of valid quantized values, the outputs which are connected to the second inputs of the quantization element, the first, second, third and fourth outputs of which form respectively the first, second, third and fourth output the quantization node, forming on the second output a second vector of quantized estimates of the modified modulation symbols, the first and second vectors have the smallest quantization error values, and at the third and fourth outputs, respectively, forming two quantization error vector values, the second, third and fourth outputs of the quantization element respectively, the second, third and fourth outputs of the quantization node, a node for generating candidate vectors, the first input of which is connected n with the second output of the channel matrix transform node, the second, third, fourth and fifth inputs of the candidate vector forming unit are connected respectively to the first, second, third and fourth outputs of the quantization node, the output of the candidate vector forming unit forming the candidate vectors output K, each of which is a set of estimates of N modulation symbols corresponding to N transmitted information signals, is connected to the first input of the binary symbol estimation node, the second input of which is connected is dined to the first input of the demodulation block, the third and fourth inputs of the binary symbol estimation node are connected to the second and third inputs of the demodulation block, the second output of the binary symbol estimation node, which forms soft estimates of the binary symbols of the information message at the second output, each of which represents a measure of the reliability of the estimate corresponding binary symbol, is the second output of the demodulation block, which is connected to the second input of the decision block. 21. A device for receiving a multicomponent signal in a radio communication system with N transmission channels and M reception channels, containing M antennas and, accordingly, M reception channels for receiving a multicomponent signal, a channel estimator, a demodulation unit, and a decision block, while the inputs of the M antennas are signal inputs devices, and the outputs are connected respectively to the inputs of the M receiving channels, the outputs of which are connected via a data bus to the input of the channel estimator and the first input of the demodulation unit, the inputs of which are signal inputs and, the second and third inputs of the demodulation block are connected respectively to the first and second outputs of the channel estimator, which forms at the first output an estimate signal for the complex transmission coefficients of the multicomponent signal during propagation from each of the transmitting antennas to each of the receive antennas, and at the second output, an estimate signal the noise dispersion of the reception channels, the first output of the demodulation block, which forms a hard estimate of the binary symbols of the information message at the first output, is connected to the first input of the decision block the output of which is the output of the device, the demodulation unit contains a node for generating a vector of the received signal, a node for converting the channel matrix, a node for evaluating the modified modulation symbols, a quantization node quantizing the vector elements

estimates of the modified modulation symbols and generating at the first output a first vector of quantized estimates of the modified modulation symbols and a binary symbol estimation node, wherein the input of the received signal vector generation section is the first input of the demodulation block, the first and second inputs of the channel matrix transform node are second and third, respectively the inputs of the demodulation unit, the output of the received signal vector forming unit is connected to the first input of the modified modulation symbol evaluation unit the second input of which is connected to the first output of the channel matrix transform node, the output of the modified modulation symbol estimation node is connected to the first input of the quantization node, the second input of which is connected to the second output of the channel matrix transform node, the first output of the binary symbol estimation node is the first output of the demodulation block, characterized in that the quantization unit comprises a memory element, a shear vector generator, a linear transformation element and a rounding element, wherein the first input is formed dividing the shear vector is the second input of the quantization node, the second input of the shear vector generator is connected to the first output of the memory element, the second output of which is connected to the first input of the linear transformation element, the second input of which is connected to the output of the shear vector generator, the third input of the linear transformation element is the first input the quantization node, the output of the linear transform element is connected to the input of the rounding element, the first, second, third and fourth outputs of the rounding element form accordingly, the first, second, third and fourth outputs of the quantization node, forming on the second output a second vector of quantized estimates of the modified modulation symbols, the first and second vectors have the smallest quantization error values, and at the third and fourth outputs, respectively generating two quantization error vector values, in the demodulation unit introduced the node forming the candidate vectors, the first input of which is connected to the second output of the node transformation channel matrix, second, third, fourth the fourth and fifth inputs of the candidate vector formation unit are connected respectively to the first, second, third and fourth outputs of the quantization node, the output of the candidate vector formation unit, which generates candidate candidates at the output K, each of which is a set of estimates of N modulation symbols corresponding to N transmitted information signals, connected to the first input of the evaluation node of binary symbols, the second input of which is connected to the first input of the demodulation unit, the third and fourth inputs of the evaluation node d of different symbols are connected respectively to the second and third inputs of the demodulation block, the second output of the binary symbol estimation unit, which forms soft estimates of the binary symbols of the information message at the second output, each of which represents a measure of the reliability of the evaluation of the corresponding binary symbol, is the second output of the demodulation block, which is connected to the second input of the decision block.

Images