KR20080042383A - Apparatus and method for receving transmitted signal in multiple antenna system - Google Patents

Abstract

An apparatus and a method for receiving a signal in a multiple antenna system are provided to perform LR-SIC based decoding in the multiple antenna system, thereby reducing the amount of decoding calculation and realizing a decoder close to the optimal BER(Bit Error Rate) performance. An SIC(Successive Interference Cancellation) unit(203) removes an LR(Lattice-Reduction) coded bit signal previously decoded from an LR receiving signal according to a predetermined rule. A signal mapping unit(205) performs the signal mapping of a signal outputted from the SIC unit by using a normalized signal mapping function or a specified signal mapping function. A decoding unit(207) receives the mapping signal to generate an LLR(Log Likelihood Ratio) for an LR information bit and an LLR for an LR coded bit and output the LR coded bit and the LR information bit. A signal reconfiguration unit(215) reconfigures the LR coded bit as a signal and feeds back the reconfigured signal to the SIC unit.

Description

Receiver and method in multi-antenna system {APPARATUS AND METHOD FOR RECEVING TRANSMITTED SIGNAL IN MULTIPLE ANTENNA SYSTEM}

1A is a block diagram of a transmitter of a multi-antenna communication system according to the present invention;

1B is a block diagram of a receiver of a multi-antenna communication system according to the present invention;

2 is a detailed receiver configuration of a multi-antenna communication system according to an embodiment of the present invention;

3 is a flowchart illustrating a receiver of a multi-antenna communication system according to an exemplary embodiment of the present invention;

4 is a mapping graph for converting a channel decoding input signal according to the present invention;

5 is a graph comparing BER performance in a multi-antenna system according to an exemplary embodiment of the present invention;

6 is a graph comparing FER performance in an orthogonal frequency division multiplexing (OFDM) system according to an embodiment of the present invention;

7 is a graph comparing FER performance in an orthogonal frequency division multiplexing (OFDM) system according to another embodiment of the present invention;

8 is a graph comparing FER performance in an orthogonal frequency division multiplexing (OFDM) system according to another embodiment of the present invention.

The present invention relates to an apparatus and a method for improving the reception performance of a multi-antenna system, and more particularly, to an apparatus for decoding a combustion interference canceled received signal using a generalized signal mapping function or a specified signal mapping function. And to a method.

The multi-antenna system can perform high-speed data transmission by transmitting and receiving signals using a plurality of antennas at both the transmitting end and the receiving end. Here, since independent symbols are simultaneously transmitted through each antenna of the transmitting end of the multi-antenna system, a technique of effectively decoding the independent symbols at the receiving end has a great effect on system performance.

As the decoding technique, there is a maximum likelihood decoding technique. The maximum likelihood decoding technique converts a received signal into a log-likelihood ratio (LLR) for each coded bit while taking into account mutual effects between transmission symbols. Here, if all possible combinations of symbols are taken into account, an accurate LLR can be obtained. However, when a large number of symbols are simultaneously transmitted, the number of symbol combinations increases exponentially, making it difficult to decode in consideration of all combinations. As a method for improving this, a method of obtaining an LLR by combining only a predetermined symbol using a sphere decoder (US 2003-389690: Spherical decoder for wireless communications, LM Daivis, D. Garrett, BMHochwald, S. Ten Brink). However, even in the case of using the old decoder, it is generally necessary to consider dozens or more of symbol combinations, which causes a large increase in the calculation amount of the entire decoding algorithm.

Therefore, many suboptimal algorithms have been studied to reduce the complexity of the decoding algorithm. For example, successive interference cancellation (hereinafter referred to as " SIC ") is a method of decoding with low computational capacity. The SIC sequentially decodes several symbols transmitted simultaneously and removes the decoded signal components from the received signal. By doing so, the process of considering symbol combinations can be omitted by separating and decoding symbols transmitted simultaneously. The SIC has an advantage of reducing the amount of calculation compared to the maximum likelihood decoding algorithm, but has a disadvantage of relatively poor performance. In particular, since the channels between antennas are similar to each other, performance deterioration is further increased in a high correlation environment.

As a method of increasing the performance of a sub-optimal algorithm such as the SIC, lattice reduction (hereinafter referred to as "LR") and a decoding technique combined with the SIC may be considered. The LR is a method of simultaneously obtaining a low complexity and high performance by applying a suboptimal algorithm such as SIC after converting the channel matrix so that the channel matrix is as orthogonal as possible. Therefore, the LR-SIC combining the LR and the SIC shows a high performance very close to the maximum likelihood decoding in a channel uncoded system. However, no effective decoding algorithm based on LR-SIC is proposed in channel coded systems. In other words, in the prior art, the LR-SIC sequentially decodes transmission signals to be decoded, and then feeds back the decoded transmission signals to remove them from the received signals. Then, decoding of the next line is performed again. At this time, if an error occurs in the process of decoding the received signal, the received signal of the wrong value is removed, resulting in performance degradation in the subsequent decoding process. This is called error propagation, which is considered to be the main cause of performance degradation.

As described above, the prior art requires a high calculation amount that is exponentially proportional to the number of transmit antennas for the maximum likelihood decoding algorithm. In the case of SIC, the computational deterioration is large in a small amount of calculation but high channel correlation. In the case of LR-SIC, no effective decoding algorithm is proposed for the channel coded system.

Therefore, there is a need for a decoder and method having a low complexity and high performance based on LR-SIC for a channel coded multi-antenna system.

Accordingly, an object of the present invention is to provide an apparatus and method for improving decoding performance based on LR-SIC in a multi-antenna system.

Another object of the present invention is to provide an apparatus and method for improving reception performance by decoding a transmission signal to which noise is added when decoding based on LR-SIC in a multi-antenna system.

An apparatus according to the present invention for achieving the above object, in a receiving apparatus in a multi-antenna system, to remove the signal of the LR coded bits previously decoded from the Lattice reduction (LR) received signal by a predetermined rule A continuous signal canceling unit, a signal mapping unit for mapping a signal output from the continuous interference removing unit using a generalized signal mapping function or a specified signal mapping function, and receiving the mapped signal for the LR information bit A decoder for generating a log likelihood ratio and a log likelihood ratio for LR coded bits to output the coded bits and the information bits, and reconstructing the LR coded bits into signals to feed back to the continuous interference canceller. Characterized in that it comprises a signal reconstruction unit.

A method according to the present invention for achieving the above object, in a receiving apparatus in a multi-antenna system, to remove the signal of the previously decoded LR coded bits in the Lattice reduction (LR) received signal by a predetermined rule A process of mapping a signal output from the continuous interference canceller using a generalized signal mapping function or a specified signal mapping function, and receiving the mapped signal and a log likelihood ratio for an LR information bit. Likelihood Ratio) and generating a log likelihood ratio for the LR coded bits, outputting the coded bits and the information bits, and reconstructing the LR coded bits into signals to feed back to the continuous interference canceller. It features.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. In describing the present invention, when it is determined that a detailed description of a related known function or configuration may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted.

Hereinafter, an apparatus and method for decoding using a generalized signal mapping function and a specified signal mapping function in an LR-SIC based decoding process in a multi-antenna system will be described.

1 shows a transceiver configuration of a multi-antenna system.

FIG. 1A illustrates a transmitter of a multi-antenna system and includes an encoder 100, a modulator 102, a demultiplexer 104, and a plurality of transmit antennas.

The encoder 100 encodes an input bit string at a corresponding code rate in order to control an error occurring on a channel, and outputs coded bits or symbols to the modulator 104. Here, when the number of input information bits is k and the code rate is R, the number of output symbols is k / R. For example, the encoder 100 may be implemented as a convolutional encoder, a turbo encoder, a low density parity check (LDPC) encoder, or the like.

The modulator 102 outputs the encoded data from the encoder 104 into complex symbols according to a corresponding modulation scheme. For example, in the modulation scheme, Binary Phase Shift Keying (BPSK), which maps one bit to one deep symbol (complex symbol), and 8-BPSK (8-ary Phase), which maps three bits to one complex symbol. Shift Keying, 16QAM for mapping four bits to one complex symbol, 64QAM for mapping six bits to one complex symbol, and the like.

The demultiplexer 104 demultiplexes the parallel complex symbols provided from the modulator 102 and transmits them through the transmit antennas. For example, when Orthogonal Frequency Division Multiple (OFDM) is used, a plurality of streams output from the demultiplexer 104 are OFDM modulated. Subsequently, the OFDM modulated signal is RF-processed to enable transmission on an actual radio channel and then transmitted to the radio channel through a corresponding antenna.

FIG. 1B illustrates a receiver of a multi-antenna system, and includes a plurality of receiving antennas, a detector 101, a decoder 103, and the like.

The detector 101 converts a signal received through the receiving antennas into an LR received signal through an LR process. Subsequently, the signal of the previously decoded coded bit is removed from the LR signal to perform signal mapping with a decoded input signal, and then output to the decoder 103.

The LR received signal is composed of an orthogonal channel matrix represented by a product of a channel matrix and a transform matrix orthogonalizing the channel matrix, and an LR transmission signal represented by a product of an inverse matrix of the transform matrix and a transmission signal matrix.

The reason why the detector 101 directly uses the added LR transmission signal, which is noise, is to improve performance by reducing error propagation due to error compensation of decoding.

However, since the LR transmission signal to which the noise to be decoded is added is not a codeword, it cannot be directly used as a decoding input. Therefore, in order to use the noise-added LR transmission signal as the input of the decoder 103, the detector 101 has noisy signals through signal mapping. The added LR transmission signal is converted into a code word.

In other words, if the sum of two codewords in the Galois field (GF) becomes another valid codeword, the LR transmission signal can be converted into codewords through signal mapping.

Although various detection techniques may be used for the detector 101, the present invention uses a method of detecting successive interference cancellation (SIC) based on lattice reduction.

The decoder 103 receives a signal detected and detected from the detector 101 (a signal in which noise is added to a symbol transmitted from a transmitter), and a log likelihood ratio (LLR) value for information bits and a log for encoded bits. A hard decision of the likelihood ratio (LLR) value outputs an information bit. In addition, the hard decision value for the estimated coded bit is fed back to the detector 101.

2 illustrates a receiver configuration of a multiple antenna system according to another exemplary embodiment of the present invention. This shows the receiver structure of FIG. 1 (b) in detail.

To help understand the description of FIG. 2, the received signal model will be described first.

개의 송신안테나를 통해 송신한 신호를 상기 도 1(b)의 수신기에서

개의 수신안테나를 사용하여 수신한 복소수 수신 신호는 하기 <수학식 1>로 표현된다.In the transmitter shown in FIG.

In the receiver of FIG. 1 (b), a signal transmitted through two transmission antennas is transmitted.

A complex received signal received using two receive antennas is represented by Equation 1 below.

는

크기의 복소 수신신호 행렬이며,

는

크기의 복소 송신 심벌 행렬이며,

는

크기의 복소 채널행렬로서,

의 심벌 시간 동안의 채널 변화량을 무시할 수 있다고 가정한다.

는

크기의 복소 잡음신호 행렬이다. 상기

는 송신안테나 개수,

은 수신안테나 개수,

은 데이터 심벌구간이다. 상기

각 송신 심벌은 BPSK나 QPSK로 변조(modulation)되어 있다고 가정하고. 이후 설명의 편의를 위해 QPSK인 경우를 중심으로 설명하기로 한다.here,

Is

Is a complex received signal matrix of size,

Is

Is a complex transmit symbol matrix of size

Is

Complex channel matrix of size

It is assumed that the channel variation during the symbol time of can be ignored.

Is

It is a complex noise signal matrix of magnitude. remind

Is the number of transmit antennas,

Is the number of antennas received,

Is the data symbol interval. remind

Assume that each transmit symbol is modulated with BPSK or QPSK. For convenience of explanation, the following description will focus on the case of QPSK.

Thereafter, the complex signal expression is converted into a real expression as shown in Equation 2 for convenience of expression development.

는 실수 수신신호 행렬, 상기

는 실수 채널행렬, 상기

는 실수 송신신호 행렬, 상기

는 실수 잡음신 호 행렬이다. 즉, 행렬들은 실수 원소를 가지며, 원래 복소수 행렬에서 크기가 확장된 형태이다. 예를 들어

는

의 크기를 가진다. 상기

는 송신안테나 개수,

은 수신안테나 개수이다.Where

Is a real received signal matrix,

Is the real channel matrix,

Is a real transmission signal matrix,

Is the real noise signal matrix. That is, the matrices have real elements and are expanded in size in the original complex matrix. E.g

Is

Has the size of. remind

Is the number of transmit antennas,

Is the number of receiving antennas.

는 하기 <수학식 3> 같이 표현된다.Where

Is expressed as Equation 3 below.

는

의 각 행으로서 서로 독립된 부호어(codeword)를 의미한다.

는 전치 행렬(transpose)를 의미한다. 여기서 QPSK 변조를 가정했으므로

의 각 원소는 1 또는 -1의 값을 가진다. 하지만 이를 0과 1을 값으로 가지는 새로운 행렬

로 변환한 송신신호는 하기 <수학식 4> 으로 표현할 수 있다.

Is

Each of the lines represents independent codewords.

Means transpose. I assumed QPSK modulation here

Each element of has a value of 1 or -1. But this is a new matrix with 0 and 1 as values

The transmission signal converted into can be expressed by Equation 4 below.

는 변환된 수신신호 행렬, 상기

는 변환된 채널행렬, 상기

는 변환된 송신신호 행렬, 상기

는 변환된 잡음신호 행렬이다.

은 모든 원소가 1인

벡터로 정의하며,

의

번째 행을

로 표시하기로 한다.Where

Is the transformed received signal matrix,

Is the converted channel matrix,

Is a transformed transmission signal matrix,

Is the transformed noise signal matrix.

Is any element of 1

Defined as a vector,

of

The first row

It is indicated by.

의 각 행들이 서로 직교성을 가지도록 하기 위해서 상기 채널행렬

을 직교성을 가지는 채널행렬

로 변환시킨다. 변환행렬을

라고 표시할 때, 상기 변환은 LR부(201)가 상기 변환행렬

을 구한 후 상기 채널행렬

과 곱하여 이루어진다. 상기

와 상기

의역행렬

는 정수로만 이루어진 행렬이며,

의 성질을 만족한다. 여기서, 상기 변환 행렬

는

가 최대한 직교성을 가지도록 해야 하며,

가 직교 행렬에 가까울수록 검출기는 높은 성능을 낼 수 있다. 상기 변환행렬(

)을 찾는 알고리즘으로 Lenstra, Lenstra and Lovasz (LLL) 알고리즘, Seysen 알고리즘 등이 있다. Referring to FIG. 2, the LR unit 201 outputs a continuous interference cancellation unit 202 by lattice reduction of the received signal represented by Equation 4 through the reception antenna. In other words, the LR unit 201 is a channel matrix of the received signal components.

The channel matrix in order to make each row of orthogonal to each other

Channel matrix with orthogonality

To. Transform matrix

, The conversion is performed by the LR unit 201 in the conversion matrix.

Find the channel matrix after

Is multiplied by remind

And above

Paraphrase

Is a matrix of only integers,

Satisfies the nature of. Where the transformation matrix

Is

Should be as orthogonal as possible,

The closer to the orthogonal matrix, the higher the detector's performance. The transformation matrix (

Algorithms to find) are Lenstra, Lenstra and Lovasz (LLL) algorithm, and Seysen algorithm.

를 하기 <수학식 5>로 변형하는 것이다.In summary, the LR unit 201 receives the received signal of Equation 4.

To transform to Equation 5 below.

는 송신신호 행렬이고, 상기

는 직교성을 가지는 채널행렬로 채널행렬과 변환행렬의 곱(

)으로 구해진다. 상기

는 변환행렬의 역행렬과 송신신호 행렬의 곱(

)으로 송신신호의 변형된 행렬이다. 여기서 상기

의

번째 행을

로 표시하기로 한다. 상기

과 상기

가 정수로 이루어진 행렬이므로

도 정수 값의 원소를 갖는다.Where

Is a transmission signal matrix,

Is an orthogonal channel matrix that is the product of the channel matrix and the transformation matrix (

Is obtained. remind

Is the product of the inverse of the transform matrix and the transmitted signal matrix (

) Is a transformed matrix of transmitted signals. Where above

of

The first row

It is indicated by. remind

And said

Since is a matrix of integers

Degrees have elements of integer values.

를 제거한 후 행단위로 복호화할 송신신호를 신호매핑부로(205)로 출력한다. 여기서, 상기

은 채널행렬

의

번째 열에 해당하고 상기

은 복호부(207)에서 바로 이전에 추정된 부호화 비트 신호이다.The continuous interference canceller 203 is a coded bit signal received from the signal reconstruction unit 215 in the received signal from Equation 5 from the LR unit 201.

After removing, the transmission signal to be decoded in units of rows is output to the signal mapping unit 205. Where

Is the channel matrix

of

Corresponds to the second column and above

Is a coded bit signal estimated just before by the decoder 207.

에서

번째를 제외한 모든 신호성분에 가중치(weight)를 곱하는 무효화(nulling) 과정을 통해 신호

로 출력된다. 여기서, 상기

은 원래 의

번째 신호에 잡음이 추가된 신호이다. 수신단에서 이상적인 경우 송신신호 행렬

을 처리하겠지만 실제로 상기 송신신호 행렬

에 잡음이 더해진 송신신호 행렬

를 처리한다.Here, the output of the signal to be decoded in units of rows is a transmission signal matrix.

in

Signal through a nulling process that multiplies the weight of all signal components except the first

Is output. Where

Is the original of

The noise is added to the first signal. Transmit Signal Matrix in Ideal Case at Receiver

Will actually process the transmitted signal matrix

Transmitted signal matrix with noise

To process

를 신호매핑을 통해 채널 복호부(207)로

신호로 매핑하여 출력한다. 여기서, 상기

은 부호화 신호가 아니기 때문에 복호부(207)의 입력신호로 바로 사용할 수 없다. 따라서, 상기

를 상기 복호부(207)의 입력신호로 사용되기 위해서는 신호매핑을 통해 부호화 신호가 되는

으로 매핑해야 복호부(207)의 입력으로 사용가능하다.The signal mapping unit 205 receives a signal input from the interference nulling unit 205 to generate an input value of the decoding unit 207.

To the channel decoder 207 through signal mapping.

Map and output the signal. Where

Since is not an encoded signal, it cannot be directly used as an input signal of the decoder 207. Thus, the above

In order to be used as an input signal of the decoder 207, a coded signal is obtained through signal mapping.

It must be mapped to be usable as the input of the decoder 207.

)에 따라 하기 도 4(a,b,c)의

과

의 매핑함수 그래프를 사용한다.For example, in the signal mapping, the signal mapping may include a specific transformation matrix (

According to Figure 4 (a, b, c)

and

Use the mapping function graph in.

4 illustrates a mapping graph for converting a channel decoder input signal according to the present invention.

과

은 각각

과

의

번째 원소를 의미한다.

이 1.3인 경우를 예로 들어 복호 과정을 설명하면 다음과 같다. 먼저

은 0.7이 됨을 알 수 있고, 이와 같이 결정된 전체 신호

을 채널 복호화기 입력으로 사용하여 복호를 수행한다. 이때

의 복호값은 0 또는 1이 될 수 있다. 먼저 0인 경우

이 0인 것으로 복호한 것이므로

을 짝수 중 하나의 값으로 결정 내려야 한다. 여기서

은

의

번째 원소를 의미한다. 짝수 중에서

에 가장 가까운 값으로

의 복호 값

을 결정하는 것이 최소 오류를 가지는 방법이 되며, 따라서

가 된다. 반대로

의 복호값이 1인 경우는 홀수 중 하나를 골라야 하며 1.3에 가장 가까운 홀수, 즉

임 됨을 알 수 있다. 이와 같이 modulo 연산에 기반하여 0, 1이 아닌 임의의 정수값을 가지는

을 채널 복호화기를 통하여 복호할 수 있고, 복호된 신호 성분을 순차적으로 제거함으로써 오차의 전파(error propagation)을 줄일 수 있다. Referring to FIG. 4A,

and

Are each

and

of

The second element.

In the case of 1.3, the decoding process is as follows. first

Is 0.7, and the total signal determined in this way

Decode is performed using as a channel decoder input. At this time

The decoding value of may be 0 or 1. First zero

Is decoded as 0

Must be determined as one of the even numbers. here

silver

of

The second element. Out of even

Closest to

Decoded value of

Is the way to have the least error, so

Becomes Contrary

If the decoding value of is 1, one of the odd numbers must be selected and the odd number closest to 1.3,

It can be seen that. As such, based on a modulo operation, any integer value other than 0 or 1

Can be decoded through a channel decoder, and error propagation can be reduced by sequentially removing the decoded signal components.

의 특성에 따라서 보다 특정한 맵핑 함수를 사용할 수도 있다. 예를 들어

에 대응되는

의 행이

과 같이 하나의 원소만 1이고 나머지는 0이라고 하면, 이와 같은 경우

은

의 행들의 선형 조합이 아니라, 특정한 하나의 행의 값을 그대로 취하게 된다. 따라서 modulo 과정을 거칠 필요 없이

자체가 하나의 부호어가 된다.Referring to FIG. 4B, in addition to FIG. 4A.

Depending on the nature of, you can use a more specific mapping function. E.g

Corresponding to

Row of

If only one element is 1 and the rest is 0, then

silver

It is not a linear combination of the rows of, but the value of one particular row. So without having to go through the modulo process

Itself is a codeword.

에 대응되는

의 행이

이라고 가정하면, 각 원소의 가능한 값은 -1, 0, 1이 되고, 상기 도 4 (a)을 변형하여 최적화된 매핑함수를 얻을 수 있다. 이때 원소가 -1, 0, 1중 하나의 값을 가질 확률 분포를 구하여, 매핑 함수가 이에 최적화 되도록 수정할 수도 있다.Referring to FIG. 4 (c), an example of another mapping function

Corresponding to

Row of

If possible, the possible values of each element are -1, 0, 1, and the optimized mapping function can be obtained by modifying the above-described FIG. At this time, the probability distribution of an element having one of -1, 0, and 1 may be obtained, and the mapping function may be modified to be optimized.

를 송신기의 부호방식에 따라 복호화하기 위해 각 정보비트에 대한 로그 우도비(Log likelihood Ratio:LLR)값을 경판정부(209)로 출력한다. 또한, 부호화 비트에 대한 로그 우도비 값을 경판정부(211)로 출력한다. 여기서, 상기 채널 복호부(207)는 SISO-MAP, Viterbi 복호화기, 터보 복호화기 등의 채널 부호에 적합한 알고리즘을 사용하면 된다.The channel decoder 207 receives the signal received from the signal mapping unit 205.

In order to decode the signal according to the coding method of the transmitter, a log likelihood ratio (LLR) value for each information bit is output to the hard decision unit 209. The log likelihood ratio value for the encoded bit is also output to the hard decision unit 211. Here, the channel decoder 207 may use an algorithm suitable for channel codes such as SISO-MAP, Viterbi decoder, and turbo decoder.

을 결정한 후 LR 역변환부(211)로 출력한다.The hard decision unit 209 performs hard decision on the LLR value of the information bit from the decoder 207 to estimate the value.

After determining the output to the LR inverse transform unit 211.

을 입력받아 LR 공간의 정보비트를 원래 정보비트

로 역변환시킨다. 역변환은

으로 이루어진다. 여기서,

는 상기 LR부(201)에서 채널행렬

를 직교성을 가지는 채널행렬

로 변환해주는 변환행렬이다.The LR inverse transform unit 211 determines a value determined from the hard decision unit 209.

Receives the information bits in the LR space

Invert to Inverse transformation

Is done. here,

The channel matrix in the LR unit 201

Channel matrix with orthogonality

Converting matrix to.

을 결정한 후 신호재구성부(215)로 출력한다.The hard decision unit 213 hardly determines the LLR value for the coded bit from the decoder 207 to estimate the estimated value.

After determining the output to the signal reconstruction unit 215.

을 입력받아

으로 재구성한 신호를 상기 연속간섭제거부(202)로 피드백해 준다. 이는 상기 연속간섭제거부(202)에서 이전 추정된 부호화 비트 신호를 수신신호에서 제거 하기 위함이다. 여기서,

은

의

번째 열에 해당한다. 이와 같은 과정을 연속적으로 거쳐

의 모든 행에 대한 복호를 끝내고, 이를 다시 역변환 (inverse-transformation)하여 원래의 소스비트 행렬에 대한 추정치

를 얻음으로써 모든 복호 과정을 종료하게 된다. 이때 SIC의 최적 성능을 얻기 위하여 복호 순서를 단순히

으로 하지 않고, 기존에 알려진 SIC 순서화 (ordering) 알고리즘으로 복호 순서를 결정할 수 있다.The signal reconstruction unit 215 determines a value determined from the hard decision unit 213.

Take input

The reconstructed signal is fed back to the continuous interference canceller 202. This is to remove the coded bit signal previously estimated by the continuous interference canceller 202 from the received signal. here,

silver

of

Corresponds to the first column. Through this process continuously

End the decoding of all rows in and inverse-transform them again to estimate the original source bit matrix.

You can end all decoding processes by getting. At this time, the decoding sequence is simply

Instead, the decoding order may be determined using a known SIC ordering algorithm.

3 is a flowchart illustrating a receiver of a multi-antenna communication system according to another exemplary embodiment of the present invention.

을 구한 후 채널추정에 의한 채널행렬

과 곱하여 직교성이 높은 채널행렬

로 변환시키는 격자감소 알고리즘을 수행한다.Referring to FIG. 3, the receiver converts the transformation matrix in step 300.

Channel matrix by channel estimation

Channel matrix with high orthogonality

Perform a grid reduction algorithm that transforms

)이 완벽한 직교 행렬이 아니므로 간섭을 제거하여 성능 향상을 시키기 위해 수신신호

에서 피드백 받은 신호

를 제거하는 것이다.Thereafter, in step 302, the receiver receives and removes the code bit signal decoded immediately before the received signal in the LR space. This is the grid-reduced channel matrix obtained in step 300 (

) Is not a perfect orthogonal matrix, so the received signal can be

Feedback from

To remove it.

)에서 n 번째를 제외한 모든 신호성분)에 가중치(weight)를 곱하여 무효화(nulling)을 수행한다. 여기서, 다른 신호성분을 무효화함으로써, 송신신호 행렬에서 행 단위로 복호화가 가능하다.In step 304, the receiver receives another signal component (i.e., a transmission signal matrix) to process the transmission signal to be decoded row by row from the input signal from which the interference is removed.

), Nulling is performed by multiplying all signal components except nth by weight. Here, by invalidating other signal components, it is possible to decode row by row in the transmission signal matrix.

에 따라 하기 도 4a, 4b, 4c 중 하나를 선택하여 사용한다.In step 305, the receiver maps the row of the next transmission signal matrix to be decoded using a mapping function using a mapping function to use the channel encoding input value. The mapping function is a transformation matrix

According to the selected one of the following 4a, 4b, 4c used.

In step 308, the receiver receives the mapping signal and decodes the signal according to the coding method of the transmitter. That is, the log likelihood ratio (LLR) value for the information bit and the log likelihood ratio (LLR) value for the encoded bit are calculated and output as a hard decision.

을 원래 정보비트

로 역변환시킨다.Thereafter, the receiver checks whether the decoding of all the rows is finished in step 310, and if it is finished, in step 312, the receiver performs inverse transformation from the LR space to the original signal space. That is, the information bits of the hard-determined LR space

Original information bit

Invert to

If it is determined in step 308 that the decoding of all the rows is finished, the process proceeds to step 314 and reconstructs the encoded bit signal, and then proceeds to step 302.

While the application of the present invention has been described assuming a multi-antenna system up to now, it is also applicable to uplink of SDMA (space division multiple access). In the uplink of the SDMA, each terminal transmits independent codewords, and the base station simultaneously receives them. This will have the same form as the received signal assumed in the present invention, it can be seen that the decoding algorithm of the present invention can be applied without modification.

을 독립적으로 복호화기에 통과시킴으로써 채널 복호화기의 계산량을 최소로 하였다. 최대 우도 복호의 경우 각 심벌 시간마다 가능한 심벌의 조합을 따지어 LLR을 구하게 된다. 심벌 조합의 수는 송신 안테나 수에 대해 지수적으로 증가하여 이 부분이 전체적인 복호 알고리즘의 계산량을 크게 높이게 된다. 하지만 본 발명은 각

을 독립적으로 복호화기에 통과시킴으로써 이와 같이 심벌 조합을 따질 필요가 없다. 또한 최대 우도 검파는 여러 번의 반복을 거쳐 최적 성능에 도달하게 되는데, 본 발명은 이와 같은 반복이 필요 없이 한번에 복호를 수행한다. 따라서 최대 우도 복호 알고리즘과 비교했을 때 매우 낮은 계산량을 가지게 됨을 알 수 있다.Here, looking at the present invention, it can be seen that the amount of calculation is very small compared to the maximum likelihood decoding. Given the slow-changing environment of the channel, the LR does not have a significant impact on the overall computation since it only needs to be computed once initially. In the present invention, each

By independently passing through the decoder, the computation amount of the channel decoder is minimized. In the case of the maximum likelihood decoding, the LLR is obtained based on a combination of possible symbols for each symbol time. The number of symbol combinations increases exponentially with the number of transmit antennas, which greatly increases the computational complexity of the overall decoding algorithm. But the present invention

By passing the independently to the decoder, there is no need to determine the symbol combination like this. In addition, the maximum likelihood detection reaches the optimum performance through several iterations, and the present invention performs decoding at once without the need for such an iteration. Therefore, it can be seen that the calculation is very low compared to the maximum likelihood decoding algorithm.

개이고, 수신 안테나는

개인 다중 안테나 시스템에서, 복호화 계산의 복잡성을 비교하기 위해 각 심벌 시간마다 실수 곱셈(RM)과 실수 덧셈(RA) 계산량을 비교한 것 이다. 이때 LR 기반 복호 알고리즘의 계산량이 최대 우도에 비해 약 4배 적은 것을 확인할 수 있다. 상기 최대 우도 복호에서 여러 번의 반복(iteration)을 할 경우 더욱 큰 계산량 이득을 기대할 수 있다.Table 1 shows the transmission antennas in order to compare the reception performance of the present invention.

Dog, and the receiving antenna

In a personal multi-antenna system, the real multiplication (RM) and real addition (RA) calculation amounts are compared for each symbol time to compare the complexity of the decoding calculation. In this case, it can be seen that the calculation amount of the LR-based decoding algorithm is about 4 times less than the maximum likelihood. More iteration gain can be expected when multiple iterations are performed in the maximum likelihood decoding.

Table 1 compares the calculation amounts of the decoding algorithm and the maximum likelihood decoding algorithm based on LR, and FIG. 5 is a graph comparing the BER performance of the decoding algorithm and the decoding algorithm of the maximum likelihood based on LR. It is shown.

) 가정하였다. 채널 부호로 제한 길이(constraint length)가 3인 1/2 귀환 구조적 컨벌루션 부호(Recursive Systematic Convolutional:RSC) 부호를 사용하였다. 채널 복호 알고리즘으로는 log-MAP를 사용하였고, 최대 우도 복호에서는 최적 성능을 위하여 LLR을 계산할 때 모든 심벌 조합을 다 고려하도록 하였다. 반면, LR을 위해 LLL 알고리즘은 채널행렬

의 각 원소는 평균이 0이고 분산이 1인 독립(적)이며 같은 분포를 따르는(independent and identically distributed:i.i.d) 복소 정규 분포 확률 변수(Gaussian-distributed random variable)로 가정하였다. SIC를 적용할 때 MMSE(minimum mean squared error) 무효화 가중치(weight)와 순서화 알고리즘을 사용하였다.The simulation environment of FIG. 5 uses QPSK modulation to see the performance of a fixed value channel.

) Assumed. As a channel code, a 1/2 recursive system convolutional code with a constraint length of 3 was used. Log-MAP is used as the channel decoding algorithm, and in the maximum likelihood decoding, all symbol combinations are considered when calculating the LLR for optimal performance. On the other hand, for LR the LLL algorithm uses a channel matrix

Each element of is assumed to be a Gaussian-distributed random variable that is independent and identically distributed (iid) with a mean of 0 and a variance of 1. When applying SIC, the minimum mean squared error (MMSE) invalidation weight and ordering algorithm were used.

5 is a graph illustrating a result of comparing BER according to an embodiment of the present invention.

Referring to FIG. 5, in a multi-antenna system having two transmit antennas, two receive antennas, or four transmit antennas and four receive antennas, it is understood that the maximum likelihood (ML) decoding improves through several iterations. The LR-based SIC (LR-SIC) decoding algorithm of the present invention has a bit error rate (BER) performance that is very close to the maximum likelihood decoding even with a low calculation amount.

)의 평균값에 적용하여 한 부호어에 대하여 한 번만

를 계산하였으며,

와

를 곱하여

를 생성하는 것은 매 심벌시간마다 수행한다.In order to compare the performance in a more realistic environment, we simulated an orthogonal frequency division multiplexing (OFDM) system. In a multi-antenna system with 4 transmit antennas and 4 receive antennas, 1024 FFTs are used, and 864 subcarriers are used as data and pilot subcarriers. One subchannel consists of 54 subcarriers, of which 42 are assumed to be data subcarriers. One codeword is composed of six subchannels, and thus, a total of 252 data subcarriers. The channels between antennas are assumed to be independent of each other without correlation, and each channel is generated by an ITU-R channel model. Like the simulations described above, QPSK modulation is used, and MMSE invalidation weights and ordering algorithms are applied in SIC. Also, we used LLL algorithm and log-MAP algorithm. Channel matrix that changes the LLL algorithm according to the symbol time

Only once for one codeword, applied to the average of

Was calculated,

Wow

Multiply by

Generating is performed every symbol time.

6 is a graph illustrating a comparison of FER performance in an orthogonal frequency division multiplexing (OFDM) system according to an exemplary embodiment of the present invention.

에서 LR 기반 SIC 수신기가 3번 반복(iteration)을 취한 최대 우도 복호기와 약 0.5 dB의 SNR 성능 차이를 가지는 것을 확인할 수 있다.6 is a graph comparing frame error rates (hereinafter, referred to as "FER") in a Pedestrian A, 3 km / h channel.

It can be seen that the LR-based SIC receiver has an SNR performance difference of about 0.5 dB from the maximum likelihood decoder that performed three iterations.

7 is a graph illustrating a comparison of FER performance in an orthogonal frequency division multiplexing (OFDM) system according to another embodiment of the present invention.

행렬이 프레임내 어떤 채널에 대해서는 최적값이 아닐 수 있다. 이에 따라 LR 기반 SIC 수신기의 성능이 열화 되어 3번 반복을 취한 최우도 복호기와 비교했을 때 약 1 dB 정도 SNR성능이 나쁜 것을 확인할 수 있다. 하지만 SNR이 증가함에 따라 FER이 감소하는 기울기는 최우도 복호기와 동일하다. 상기 도 5과 비교했을 때 최우도 복호기의 성능이 약간 상승한 것을 볼 수 있는데, 이는 채널이 빠르게 변화함에 따라 시간 다이버시티(diversity) 효과를 얻었기 때문이다.7 is a graph of FER comparison for Pedestrian B, 10 km / h channel. As the channel changes faster, the amount of channel change cannot be ignored within a frame. Therefore, the average value of the channel

The matrix may not be optimal for any channel in the frame. As a result, the performance of the LR-based SIC receiver is degraded, and the SNR performance is about 1 dB lower than that of the maximum likelihood decoder which takes three iterations. However, as the SNR increases, the slope of FER decreases is the same as that of the maximum likelihood decoder. Compared with FIG. 5, the performance of the highest likelihood decoder is slightly increased because the time diversity effect is obtained as the channel changes rapidly.

8 is a graph illustrating a comparison of FER performance in an orthogonal frequency division multiplexing (OFDM) system according to another embodiment of the present invention.

8 is a FER comparison graph for the Vehicular A, 60 km / h channel. As the speed is further increased, the channel change is severe, and it can be seen that the LR-based SIC receiver exhibits a performance degradation of about 2 dB.

Meanwhile, in the detailed description of the present invention, specific embodiments have been described, but various modifications are possible without departing from the scope of the present invention. Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined not only by the scope of the following claims, but also by the equivalents of the claims.

As described above, by performing the LR-SIC-based decoding of the multi-antenna system, it is possible to reduce the decoding calculation amount and implement a decoder that approximates the optimal BER performance.

Claims (17)

In the receiver in a multi-antenna system, A continuous interference cancellation unit for removing a signal of a previously decoded LR coded bit from a lattice reduction received signal by a predetermined rule; A signal mapping unit for mapping a signal output from the continuous interference canceling unit by using a generalized signal mapping function or a specified signal mapping function; A decoder configured to receive the mapped signal and generate a log likelihood ratio for LR information bits and a log likelihood ratio for LR coded bits to output the encoded bits and the information bits; And a signal reconstructing unit for reconstructing the LR coded bits into a signal and feeding back the continuous interference canceling unit. The method of claim 1, And an LR inverse transform unit which inversely converts the LR information bits into original information bits. The method of claim 1, The trellis-reduced received signal is a trellis-reduced transmission signal converted into a product of an orthogonal channel matrix converted into a product of a channel matrix by channel estimation and a transform matrix orthogonalizing the channel matrix, and a product of an inverse matrix of the transform matrix and a transmission signal matrix. A device characterized by being the product of a matrix. The method of claim 1, The decoding unit is characterized in that the SISO-MAP, Viterbi decoder, turbo decoder, etc. can be applied. The method of claim 1, And the mapped signal becomes a codeword that can be used as an input of a decoder. The method of claim 5, And the mapped signal as the codeword is converted into a form in which noise is added to a binary signal of 0 and 1 in a form in which a noise is added to the LR information bits. The method of claim 1, Wherein the generalized signal mapping function is a of FIG. The method of claim 1, The specified signal mapping function is determined by a conversion matrix for converting a channel matrix into an orthogonal channel matrix. The method of claim 8, Wherein the specified signal mapping function is b, c of FIG. In the receiver in a multi-antenna system, Removing a signal of a previously decoded LR coded bit from a lattice reduction received signal by a predetermined rule; Signal mapping the signal output from the continuous interference canceller using a generalized signal mapping function or a specialized signal mapping function; Receiving the mapped signal and generating a log likelihood ratio for LR information bits and a log likelihood ratio for LR coded bits to output the encoded bits and the information bits; And reconstructing the LR coded bits into a signal and feeding the feedback back to the continuous interference canceller. The method of claim 10, And inversely converting the LR information bits into original information bits. The method of claim 10, The trellis-reduced received signal is a trellis-reduced transmission signal converted into a product of an orthogonal channel matrix converted into a product of a channel matrix by channel estimation and a transform matrix orthogonalizing the channel matrix, and a product of an inverse matrix of the transform matrix and a transmission signal matrix. Characterized by the product of a matrix. The method of claim 10, The mapped signal is a codeword that can be used as an input of a decoder. The method of claim 13, The mapped signal, which is the codeword, converts a received signal having a noise added to the LR information bits into a noise added to a binary signal of 0 and 1. The method of claim 10, The generalized signal mapping function is a in FIG. 4. The method of claim 10, The specified signal mapping function is determined by a conversion matrix for converting a channel matrix into an orthogonal channel matrix. The method of claim 16, The specified signal mapping function is b, c in the following <4>.

Images