KR101397347B1 - A method for generating soft decision information in mimo system - Google Patents

Abstract

There is provided a method of generating soft decision information on transmission bits as an input of a channel decoder for a received signal vector y in a MIMO (Multi-Input Multi-Output) system in which the number of transmission antennas is T and the number of reception antennas is N do. One column vector among the L (1? L? T) column vectors selected in the channel matrix H is replaced with a column vector located in the Tth column from the left of the matrix H , and a column located in the Tth column In the state that the vector is selected, a permutation is determined for the remaining T-1 column vectors to generate a channel matrix H ' . QR decomposition is performed on the channel matrix H ' as H' = Q'R ' . To the received signal vector y by multiplying the Hermitian (Hermitian) matrix Q ^H of the matrix Q is obtained a modified received vector z. M candidate symbol vectors are obtained using the modified received vector z for all points on the possible M-ary constellation with a modulation index of log ₂ M.

Description

[0001] The present invention relates to a method for generating soft decision information in a MIMO system,

The present invention relates to wireless communication, and more particularly, to a method for generating soft decision information as an input to a channel decoder of a receiver in a multi-input multi-output (MIMO) system.

A multi-input multi-output (MIMO) system using multiple transmit and receive antennas can significantly increase the theoretical channel capacity without additional bandwidth compared to a system using a single transmit and receive antenna. It is getting attention. In addition, turbo codes such as turbo codes are used in conjunction with MIMO systems, which are close to the theoretical channel capacity. Therefore, systems that link MIMO systems with channel encoders have been proposed in various standards.

In order to approximate the theoretical channel capacity by using a system that concatenates channel codes with the proposed V-BLAST (Vertical Bell Labs LAyered Space-Time Architecture) system as a practical MIMO system, the input of the channel decoder from the V- A process of generating soft decision information is essential. The soft decision information as input to the channel decoder corresponds to a log-likelihood ratio (LLR) for each transmitted bit mapped by an M-ary modulated constellation. Since the computational complexity is proportional to M ^T when it is performed according to the maximum likelihood algorithm, there are a lot of constraints on real hardware implementation. On the other hand, the zero-forcing algorithm proposed for the purpose of lowering the computational complexity has a much higher performance degradation than the maximum likelihood algorithm.

Therefore, an algorithm is required which has a low computational complexity and hardly degrades performance compared to the maximum likelihood algorithm.

It is an object of the present invention to provide a method of calculating a log likelihood ratio as a channel decoder input whose computational complexity is lower than that of the maximum likelihood algorithm but whose performance approximates the maximum likelihood algorithm.

There is provided a method of generating soft decision information on transmission bits as an input of a channel decoder for a received signal vector y in a MIMO (Multi-Input Multi-Output) system in which the number of transmission antennas is T and the number of reception antennas is N do. One column vector among the L (1? L? T) column vectors selected in the channel matrix H is replaced with a column vector located in the Tth column from the left of the matrix H , and a column located in the Tth column In the state that the vector is selected, a permutation is determined for the remaining T-1 column vectors to generate a channel matrix H ' . QR decomposition is performed on the channel matrix H ' as H' = Q'R ' . To the received signal vector y by multiplying the Hermitian (Hermitian) matrix Q ^H of the matrix Q is obtained a modified received vector z. M candidate symbol vectors are obtained using the modified received vector z for all points on the possible M-ary constellation with a modulation index of log ₂ M.

As described above, according to the present invention, in a MIMO system using T transmit antennas and N receive antennas and a modulation index of a symbol transmitted from each transmit antenna is log ₂ M, a candidate symbol vector Reducing the number from M ^T to MT reduces the computational complexity. There is almost no performance degradation compared to the maximum likelihood algorithm.

The present invention proposes a method for efficiently reducing the number of candidate symbol vectors required for calculating the log likelihood ratio. The received signal model of a typical V-BLAST system is shown in Equation (1).

Where T denotes the number of transmit antennas, and N denotes the number of receive antennas. In the case of the channel matrix H , each element has a variance of 1 (E [| h _{i , j} | ² ] = 1. The complex Gaussian random variable, a white Gaussian noise vector n, is defined by E [| n _{i , j} | ² ] = σ _n ² , i = 1,2, ..., N.

When h _i as the i-th column vector of the channel matrix H given in equation 1 h and _i with each other to change the position of the _{h T, h j (j =} 1,2, ..., T, j ≠ i) Are determined according to a specific criterion. Assuming that the jth column vector of the given channel matrix H before the permutation of the column vectors is determined corresponds to the σ (j) -th order in the matrix after the order is determined, Ha in Equation 1 is H You can rewrite it like 'a' .

Let Q ' R ' be the result of performing QR decomposition on H ' when N > = T and H' rank is T in Equation (2). The results are multiplied by the Hermitian (Hermitian) matrix Q ^H of the matrix Q on both sides of the equation (1) can be written as equation (3).

에 임의의 s_l 을 대입하고 수학식 4의 y_j(j=T-1,...,1)를 계산한다.Assuming that the modulation index at each transmit antenna is log ₂ M, letting all the points on the M-ary constellation be s ₁ , s ₂ , ..., s _M ,

Substituting an arbitrary s in _l and calculates of Equation _{4 y j (j = T-} 1, ..., 1).

를 얻게 된다. 여기서 벡터 x ^T는 벡터 x의 전치 (transpose)를 나타낸다. 이 과정을 모든 가능한 s_l(l=1,...,M) 에 대하여 수행하면 M 개의 후보 심벌 벡터들

을 얻게 된다. 이렇게 얻은

들의 각 원소들의 순서는 효율적인 검출을 위해 도입된 것으로 원래 순서와 다르므로 각 후보 심벌 벡터의 원소들의 순서를 원래 순서대로 재배치한다.Subsequently, when slicing is performed for each y _j , one candidate symbol vector

. Where the vector x ^T represents the transpose of the vector x . If we perform this procedure for all possible s _l (l = 1, ..., M), M candidate symbol vectors

. So obtained

Are introduced for efficient detection and are different from the original order, so that the order of the elements of each candidate symbol vector is rearranged in the original order.

When the above process is performed on all the column vectors h _i (i = 1, ..., T) of a given channel matrix H , all MT candidate symbol vectors are obtained. By calculating Equation (5) using the MT symbol vectors, a log likelihood ratio for the lambda < th > bit can be obtained and calculated for all lambda and input to the channel decoder.

Here, A _? ¹ is a set of candidate symbol vectors corresponding to the lambda-th bit among the MT candidate symbols obtained through the process described above, and A _? ⁰ is a set of candidate symbol vectors corresponding to the? It is a set.

A number of candidate symbol vectors are needed to generate soft decision information. First, each row of the received signal model is represented by a vector and a matrix. Among the various layers, the i (i = 1, ..., T) th layer from the bottom is aligned with the bottom row and the positions of the remaining layers are determined according to a specific permutation. After performing the QR decomposition, the H 'matrix H resetting the position of the column vector in accordance with a particular permutation for a given channel matrix H in = Q'R' Q in the received vector is multiplied by the Hermitian matrix of the matrix Q ^H. For all the possible candidate symbols for the first detection layer (the layer at the bottom end) and the decision feedback equalization (DFE) detection method for the remaining T-1 transmitted symbols, M candidate symbol vectors are calculated (Where the modulation index is log ₂ M). By performing this process on all i (i = 1, ..., T), MT candidate symbol vectors can be obtained.

A method of generating a log likelihood ratio for transmission bits as an input of a channel decoder in a MIMO system in which the number of transmission antennas is T and the number of reception antennas is N is as follows. One column vector among the L (1? L? T) column vectors selected in a given channel matrix H is swapped with a column vector located in the Tth column from the left of the channel matrix H. At this time, when L < T, L column vectors can be selected based on the magnitude relation of the norms of the column vectors of the channel matrix H. Or, L of the L <T in case basis depending on the specific permutation change the position of each column vector of the channel matrix H and then the magnitude relation of the absolute value of the diagonal elements of Q'R 'to perform a QR decomposition to the matrix R' Column vectors can be selected.

The permutation is determined for the remaining T-1 column vectors in a state in which the column vector to be located in the T-th column is selected. Let H ' denote the matrix whose position of the column vectors of the channel matrix H given by the determined specific permutation is changed. At this time, the permutation can be determined on the basis of the magnitude relation of the norm of each column vector. Or, Q _T with respect to the matrix H _{-1 T} consisting of the T-1 of a column vector _-1 R _{T - 1} time obtained by performing QR decomposition, each absolute value of the diagonal elements of the matrix R _{T -1} on different permutations The permutation can be determined on the basis of the magnitude relation of.

'H a' matrix H and performs QR decomposition as Q'R = ", and by multiplying the Hermitian matrix Q ^H of the matrix Q to the reception signal vector is determined for the modified received vector z. And selects one transmitted symbol from the received vector z as a candidate symbol. At this time, a transmission symbol corresponding to the T-th element (the element at the bottom of the calculated column vector) can be selected as a candidate symbol.

And selects C (1? C? M) out of all possible constellation points that the candidate symbol may have. At this time, if C <M, C number can be selected by using the modified reception vector z .

For each point selected as the candidate symbol, interference cancellation is performed from the modified reception vector z , and then C candidate symbol vectors are obtained by applying a decision feedback (DF) detection method.

All the above steps are performed for all L (1? L? T) column vectors selected from a given channel matrix H to obtain CL candidate symbol vectors.

Hereinafter, a method for selecting efficient candidate symbol vectors for log likelihood ratio calculation for each transmission bit required as an input of a channel decoder according to a preferred embodiment of the present invention will be described with reference to the accompanying drawings.

1 is a flowchart showing a soft decision information generating method according to an embodiment of the present invention for a V-BLAST system in which the number of transmit / receive antennas is two.

First, in step 102, the reception vector y and the channel matrix H are input.

The received vector y is multiplied by the hermetian matrix Q ^H of the Q matrix obtained after performing the QR decomposition on the channel matrix H given in step 104. [ And initializes l to 1 indicating an index for each point on the M-ary constellation.

In step 108, a symbol corresponding to a ₂ is assumed to be s ₁ , a symbol corresponding to a ₁ is removed from interference by a ₂ from z _1, and then the result is normalized to r ₁₁ , do.

이라 할 때

를 하나의 후보 심벌 벡터로서 출력한다. In step 110,

When you say

As a candidate symbol vector.

In step 106, steps 108 and 110 are performed on all points on the constellation until l indicating the index becomes M.

Through the repetition of steps 102 to 110, a total of M candidate symbol vectors are obtained. If l and M are equal, step 112 is performed.

In step 112, the positions of the two column vectors of the channel matrix H are swapped with each other. Let H ' be a matrix that shifts the position of two column vectors.

In step 114, the matrix is subjected to QR decomposition for H ' . QR decomposition then allows multiplying the Hermitian matrix Q 'of ^H' matrix Q to as' the obtained matrix Q'R the received vector y.

In steps 116 to 120, the same processes as those performed in steps 106 to 110 are repeated. However, in step 108, the result detected through the division corresponds to the symbol transmitted from the first antenna, while in step 118, the symbol corresponds to the symbol transmitted from the second antenna.

FIGS. 2 and 3 illustrate a method of selecting a candidate symbol vector according to a method proposed by the present invention in a system in which a CTC (Convolutional Turbo Code) is connected to a V-BLAST system in which the number of transmit and receive antennas is two, And a frame error rate (FER) when the likelihood ratio is calculated.

2 shows the performance when the soft decision information generation method according to the present invention and the soft decision information generation method according to the maximum likelihood algorithm are used in the coded V-BLAST system using the frame length of 144 bits as the CTC (Convolutional Turbo Code) FIG. 3 is a graph illustrating a soft decision information generation method according to the present invention and a soft decision information generation method according to the maximum likelihood algorithm in a coded V-BLAST system using a CTC (Convolutional Turbo Code) with a frame length of 480 bits This is a graph comparing the performance when using the method.

The parameters used in this experiment are as follows.

- Number of transmit antennas: 2

- Number of receiving antennas: 2

- Modulation scheme: QPSK, 16-QAM, 64-QAM

Information frame length: 144 bits (FIG. 2), 480 bits (FIG. 3)

- Code rate: 1/3

- Decoding method: Max-log MAP, scaling factor 0.7 applied.

- each element of the channel matrix is i.i.d with an average of 0 and a variance of 1. is an arbitrary and identically distributed complex Gaussian random variable.

As can be seen from FIGS. 2 and 3, when the present invention is applied to a coded V-BLAST system in which both the number of transmit antennas and the number of receive antennas are two, it is confirmed that there is almost no performance deterioration as compared with the performance of the maximum likelihood algorithm have.

All of the functions described above may be performed by a processor such as a microprocessor, a controller, a microcontroller, an application specific integrated circuit (ASIC), etc. according to software or program code or the like coded to perform the function. The design, development and implementation of the above code will be apparent to those skilled in the art based on the description of the present invention.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it will be understood by those skilled in the art that various changes and modifications may be made therein without departing from the spirit and scope of the invention. You will understand. Therefore, it is intended that the present invention covers all embodiments falling within the scope of the following claims, rather than being limited to the above-described embodiments.

1 is a flowchart showing a soft decision information generating method according to an embodiment of the present invention for a V-BLAST system in which the number of transmit / receive antennas is two.

2 shows the performance when the soft decision information generation method according to the present invention and the soft decision information generation method according to the maximum likelihood algorithm are used in the coded V-BLAST system using the frame length of 144 bits as the CTC (Convolutional Turbo Code) FIG.

FIG. 3 is a flowchart illustrating a method of generating a soft decision information according to the present invention and a soft decision information generation method according to a maximum likelihood algorithm in a coded V-BLAST system using a CTC (Convolutional Turbo Code) with a frame length of 480 bits This is a graph comparing performance.

Claims (2)

A method for generating soft decision information on transmission bits as an input of a channel decoder for a received signal vector y in a multi-input multi-output (MIMO) system in which the number of transmit antennas is T and the number of receive antennas is N , One column vector among the L (1? L? T) column vectors selected in the channel matrix H is replaced with a column vector located in the Tth column from the left of the matrix H , and a column located in the Tth column Determining a permutation for the remaining T-1 column vectors with the vector selected and generating a channel matrix H ' ; Performing QR decomposition such that H ' = Q'R' for the channel matrix H ' ; Obtaining a modified received vector z to the received signal vector y by multiplying the Hermitian (Hermitian) matrix Q ^H of the matrix Q; And Determining a M candidate symbol vector using the modified received vector z for all points on a possible M-ary constellation with a modulation index of log ₂ M. The method according to claim 1, And repeating the above steps for all T column vectors of the channel matrix H to obtain MT candidate symbol vectors.

Images