KR100958319B1 - Signal detecting apparatus for Multiple-Input Multiple-Output system - Google Patents

Abstract

The present invention relates to a signal detection apparatus of a multiple input multiple output system, comprising: an input memory unit for storing an input signal; A signal generator for calculating a signal for a partial Euclidean distance (PED) operation; A PED calculator for calculating a PED value based on a symbol of a node retrieved at a predetermined level of each node and a symbol of a node retrieved at a previous level according to a signal input from the input memory unit and the signal generator; An alignment unit for selecting K PEDs calculated in the PED operation unit in order of decreasing size and searching for a corresponding pair of symbols; And a symbol pair storage unit for storing the symbol pairs found in the alignment unit. As a result, signal detection performance and processing speed may be improved, and various modulation schemes may be supported without additional hardware.

Description

Signal detecting apparatus for multiple input multiple output system

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a signal detection apparatus of a multiple input multiple output system, and more particularly, to a multiple input multiple output system for transmitting and receiving signals in a space-division multiplexing scheme. A signal detection device of a multi-input multiple output system capable of improving signal detection performance and processing speed of a signal detection device that detects spatially multiplexed data and capable of supporting various modulation schemes without adding additional hardware. will be.

As high quality and high speed data transmission is required in a wireless communication environment, a multiple-input multiple output (MIMO) system using multiple antennas is used to efficiently use a limited frequency. have.

The MIMO system may be operated by space-time coding or space-division multiplexing. Space-time encoding is a technology that can improve the reliability of a wireless communication system by encoding data transmitted from different antennas. Space-division multiplexing is a technique for increasing data rate by simultaneously transmitting data independent of each antenna.

In case of transmitting an independent symbol for each transmitting antenna through spatial division multiplexing in a MIMO system, various techniques for detecting a transmitting symbol from the receiving symbol at the receiving end have been proposed. In particular, the ML (Maximum Likelihood) detection technique is a technique that calculates and compares Euclidean distance with respect to symbol vectors in all cases that can be transmitted for symbol detection. However, as the number of antennas and the size of the modulation scheme increase, the complexity increases exponentially, which makes the implementation very difficult.

In order to reduce the complexity of such ML detection, a sphere decoding technique has been developed. The sphere decoding technique reduces the complexity of ML detection by performing Euclidean distance calculation only on a set of symbol vectors existing in a sphere having an initially set radius in consideration of noise variance and channel conditions.

In general, spear decoding techniques can be classified into depth-first search and width-first search.

The depth-first search algorithm can reduce the complexity of ML detection by tree searching back and forth. However, the digital circuit cannot be implemented by using a pipeline characteristic that can increase throughput, and since the calculation amount is variable, there is a problem that a search having a complexity of ML detection has to be performed in the worst case.

The breadth-first search algorithm, on the other hand, is designed to find the best candidate only in the forward direction. In particular, the K-best sphere decoding algorithm stores only the K best candidates at each tree search level.

FIG. 1 is a node selection state diagram according to a conventional sphere decoding method, and illustrates a process of selecting an optimal candidate group according to a K-best sphere decoding algorithm.

As shown in FIG. 1, nodes 1, 3, 5, and 7 exist at level i , and each node has a cumulative partial Euclidean distance (PED) value of 0.2, 0.3, 0.4, and 0.5. Here, node 7 is eliminated from the candidate group because the PED value is the highest as 0.5.

When calculating the new value PED selected candidate nodes are expanded to a level i-1 from the level i, and again updates the candidate node is selected nodes 2, 4, 6. Here, node 8 has a PED value of 0.55 and node 6 has a PED value of 0.7. Since node 8 has a lower PED value, but node 7 has already been discarded at level i , the best candidates for both levels of discovery are Node pairs (1, 2), (3, 4), (5, 6).

This K-best sphere decoding algorithm has a disadvantage in that if the K value is not large enough, the decoding delay is increased even though the performance degradation may be caused by the error transfer of the previous level and the throughput may be kept constant.

In addition, since the hardware structure of the conventional K-best sphere decoding algorithm scheme supports only one modulation scheme, it is necessary to add a hardware configuration in order to support various modulation schemes, thereby increasing the design area.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problem, and the present invention provides a method for spatially multiplexing data in a multiple-input multiple-output system that transmits and receives a signal in a space-division multiplexing scheme. It is an object of the present invention to provide a signal detection device of a multi-input multiple output system that can improve signal detection performance and processing speed of a signal detection device to detect and can support various modulation schemes without adding additional hardware.

According to an aspect of the present invention, there is provided a signal detecting apparatus for a multiple input multiple output system, comprising: an input memory unit for storing an input signal; A signal generator for generating a signal for calculating a signal for a partial Euclidean distance (PED) operation; A PED calculator for calculating a PED value based on a symbol of a node retrieved at a predetermined level of each node and a symbol of a node retrieved at a previous level according to a signal input from the input memory unit and the signal generator; An alignment unit for selecting K PEDs calculated in the PED operation unit in order of decreasing size and searching for a corresponding pair of symbols; And a symbol pair storage unit for storing the symbol pairs found in the alignment unit.

Here, when the PED calculation unit is at level i = M, M-1, M-2, ..., 1. and PED L _i (s ^(j-1) ) at the level i , 6], it is possible to calculate the PED value based on the symbol of the node retrieved at the previous level of the tree search and the symbol of the node retrieved at the current level.

&Quot; (6) "

는 트리 탐색의 이전 레벨에서 찾아진 심벌의 부분 벡터이다.Where j = 2i,

Is the partial vector of symbols found at the previous level of tree traversal.

Further, when the PED calculation unit is at level i = M, M-1, M-2, ..., 1. and PED L _i (s ^(j-1) ) at the level i , 7], it is possible to calculate the PED value based on the symbol of the node retrieved at the previous level of the tree search and the symbol of the node retrieved at the current level.

[Equation 7]

항과

항은 이전 레벨에서 검출된 s ^(j+1) 에 의해서만 영향을 받는다.here,

Section

The term is only affected by s ^{(j + 1)} detected at the previous level.

On the other hand, the alignment unit, and a multiplexer for selectively outputting the PED value output from the PED operation unit; A sorter for sorting and outputting the PED values inputted through the multiplexer in ascending order; Memory A and memory B for storing PED values output from the sorter; According to a modulation method of the input signal, the memory A and the memory B may include an alignment control unit for controlling to operate by a clock gating technique in which the operation is stopped when the memory logic is not used.

The sorter may include a buffer configured to receive a PED value from the multiplexer every predetermined clock cycle; It is possible to include a plurality of comparators for comparing the size of the PED value input to the buffer and output in ascending order.

In addition, the modulation scheme of the input signal may include BPSK, QPSK, 16-QAM, and 64-QAM.

Hereinafter, a signal detection apparatus of a multiple input multiple output system according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings. However, in describing the present invention, if it is determined that a detailed description of a related known function or configuration may unnecessarily obscure the subject matter of the present invention, a detailed description thereof will be omitted.

The multiple-input multiple-output (MIMO) system to which the present invention is applied consists of M transmit antennas and N receive antennas, and the baseband N-dimensional receive symbol vector is represented by 1].

는 M차원 송신 신호 벡터이고 각 신호성분들은 BPSK, QPSK, 16-QAM, 64-QAM과 같은 복소성좌도 내에서 독립적으로 나타나는 심벌이다.

는 N×M 채널 행렬이고, 행렬 내의

는 j번째 송신안테나와 i번째 수신안테나 사이의 전달함수이다. 모든

는 각 차원마다 평균값이 0이고 분산이 0.5인 i.i.d 복소 랜덤 가우시안 변수이고, 수신기에서는 이 채널 정보를 모두 알고 있다고 가정한다.

는 평균 0, 분산

인 i.i.d 복소 가우시안 잡음이다. [수학식 1]의 복소수 형태의 식은 아래의 [수학식 2]로 나타낼 수 있다.here

Is an M-dimensional transmission signal vector and each signal component is a symbol that appears independently in a complex locus such as BPSK, QPSK, 16-QAM, and 64-QAM.

Is an N × M channel matrix, and

Is a transfer function between the j th transmit antenna and the i th receive antenna. all

Is an iid complex random Gaussian variable with an average value of 0 and a variance of 0.5 for each dimension, and assume that the receiver knows all of this channel information.

Is an average of 0, variance

Iid is a complex Gaussian noise. The complex form of Equation 1 may be represented by Equation 2 below.

과

는 (ㆍ)의 실수부와 허수부이다. n을 2N, m을 2M이라 할 경우, [수학식 2]의 N×M채널 행렬

는 아래의 [수학식 3]과 같이 QR분할방식으로 삼각행렬로 표현할 수 있다.here

and

Is the real part and the imaginary part of (·). When n is 2N and m is 2M, the N × M channel matrix of Equation 2

Can be expressed as a triangular matrix by the QR division method as shown in Equation 3 below.

를 곱하면 다음의 [수학식 4]와 같이 표현할 수 있다.Where R is an n × m upper triangular matrix and Q is an n × m matrix with orthogonal columns. In [Equation 2]

By multiplying it can be expressed as Equation 4 below.

이고

이다. MIMO 검출의 목적은 성상도 상에서 가장 가까운 심벌위치,

를 찾는 것이고 이는 [수학식 5]와 같이 나타낼 수 있다.here

ego

to be. The purpose of MIMO detection is to locate the nearest symbol position on the constellation,

This can be expressed as [Equation 5].

에서 정의된다.Where the value of each s is the point Ω in real constellation

Is defined in.

2 is a control block diagram of a signal detection apparatus of a multiple input multiple output system according to an exemplary embodiment of the present invention, illustrating a 4 x 4 MIMO signal detection apparatus. Thus, K _i consists of a tree of four levels = 4, level i = 2, 3, 4, and can support up to four other modulation schemes (BPSK, QPSK, 16-QAM, 64-QAM).

As shown in FIG. 2, the signal detection apparatus of the present invention generates a signal for calculating a signal for RAM Y (2) and RAM R (4) for storing an input signal and a PED (partial Euclidean distance) operation. The PED calculator 100 calculates the PED according to the signals input from the signal generator 6, the RAM Y (2), the RAM R (4), and the signal generator 6, and the PED calculator 100 Select the optimal K PED values from the calculated PED values and store the symbol pairs in RAM PE1 (12), RAM PE2 (14), RAM PE3 (16), and RAM PE4 (18), and store the accumulated values in RAM PED ( And a control unit 10 for controlling the operation thereof.

In RAM Y (2) and RAM R (4), input signals to be detected are stored in the form of variables y and R according to [Equation 3] and [Equation 5].

The signal generator 6 calculates a potential candidate group { s _j , s _j-1 } used in the PED calculation in the PED calculator 100.

The sorter 200 sorts the PED values output from the PED calculator 100, finds the optimal K _i values, and finds a corresponding pair of symbols { s _j , s _{j -1} }.

The RAM PED 8 stores PED values accumulated after the sorting operation in the PED calculator 100.

RAM PE1 (12), RAM PE2 (14), RAM PE3 (16), and RAM PE4 (18) are symbol pairs { s ₈ , s ₇ }, { s ₆ , s ₅ }, { s ₄ , s ₃ } respectively. , { s ₂ , s ₁ }. This symbol pair is an optimal candidate group determined by the sort operation, and is output through the output buffer 18.

Here, the PED calculation unit 100 performs symbol detection at two levels at the time of tree search in order to exclude the possibility that potential candidate groups that can help improve performance in the process of finding an optimal candidate group are deleted. At each level, K _i symbol pairs are maintained.

3 is a control block diagram of the PED calculation unit, and may be configured with operation blocks for PED operation.

During signal search, the PED value is initialized by setting initial PED L _{M +1} = 0, input y, R, and the PED calculator 100 calculates the PED value at each level i. Here, when i = M, M-1, M-2, ..., 1. PED L _i (s ^(j-1) ) at the level i may be calculated as shown in [Equation 6].

는 트리 탐색의 이전 레벨에서 찾아진 심벌의 부분 벡터이다. [수학식 6]은 [수학식 7] 같이 다시 나타낼 수 있다.Where j = 2i. In [Equation 6]

Is the partial vector of symbols found at the previous level of tree traversal. Equation 6 may be represented again as in Equation 7.

항과

항은 이전 레벨에서 검출된 s ^(j+1) 에 의해서만 영향을 받는다. Ω가 Q개이고 레벨 i-1에서 K _{i -1} 개의 후보군 s ^(j+1) 을 갖는다면, 이 단계에서 전체 PED 값의 개수는 G= K _{i -1} ×Q가 된다.In [Equation 7]

Section

The term is only affected by s ^{(j + 1)} detected at the previous level. If Ω is Q and has K _{i -1} candidate groups s ^{(j + 1)} at level i-1 , the total number of PED values at this stage is G = K _{i -1} x Q.

및

등이 합산되어 최종적으로 PED L _i (s ^(j-1) )이 출력된다. PED calculation unit 100 is composed of a plurality of operation blocks that can perform the operation of the equation (7). First, signals input from the RAM R 4 and the signal generator 6 are stored in the storage blocks 110, 112, and 114. The values y _{j -1} and y _j output from the RAM Y (2) at the first sum block 120 and the second sum block 126 are added to the stored input signals, respectively. After that, the computed signal is calculated according to s ^{(j + 1} ) detected at the previous level through the summation blocks 122, 124, 128, and 130.

And

Etc. are summed and finally PED L _i (s ^(j-1) ) is output.

When searching for an optimal candidate group through the PED calculator 100 having such a configuration, symbol detection may be simultaneously performed at two levels during tree search. 4 is a node selection state diagram of the PED calculator 100 of the present signal receiving apparatus.

4, the level of the node i exists, 1, 3, 5, 7, and nodes existing in, a level i-1 2, 4, 6 , 8 which is in the search at the same time. At level i, there are nodes 1, 3, 5, and 7, and each node has a cumulative PED of 0.2, 0.3, 0.4, and 0.5. At level i-1 , nodes 2, 4, 6, and 8 exist, and each node has a cumulative PED value of 0.3, 0.4, 0.7, and 0.55.

Therefore, when the search for a node of a node and a level i-1 of level i at the same time, the node pair final cumulative PED value of (7,8) is 0.55, and the current level because the final cumulative PED 0.7 the nodes (5, 6) The finally selected candidate groups of are (7, 8), (1, 2), (3, 4).

As described above, the PED calculation unit 100 performs symbol detection at two levels at the time of tree search in order to exclude the possibility that potential candidate groups that can help improve performance in the process of finding an optimal candidate group are deleted. . At each level of the search, K _i symbol pairs are maintained. Compared with the conventional algorithm, the number of search levels of the proposed algorithm is reduced by half, and the performance is improved because the symbol search space for searching for potential candidates is expanded.

Meanwhile, the sorting unit 200 sorts the PED L _i (s ^(j-1) ) output from the PED calculating unit 100 in ascending order, finds the optimal K _i values, and corresponds to the symbol pair { s _j , s _{j −1} }, while storing the accumulated PED values in the RAM PED 8 after the sort operation. The configuration of the alignment unit 200 is as shown in FIG.

5 is a control block diagram of the alignment unit 200 according to an embodiment of the present invention, illustrating a configuration that may be implemented in the 4 x 4 MIMO signal detection apparatus of FIG.

As shown in FIG. 5, the sorter 200 sorts the PED values input through the multiplexer 210 and the multiplexer 210 in ascending order by selecting the PED values output by the PED calculator 100 one by one. And a sorter 220 for outputting the memory, memory A and B 240 and 250 for storing the PED values output from the sorter 220, and an alignment controller 230 for controlling them.

The multiplexer 210 selectively outputs the input PED values according to the selection signal SEL output from the alignment controller 230.

Aligner 220 has 16 inputs and supports four modulation schemes: BPSK, QPSK, 16-QAM, and 64-QAM. Accordingly, the sorter 220 having 256 and 64 inputs may be implemented as the sorter 220 having 16 inputs, thereby reducing complexity.

Memory A, B (240, 250) is composed of 4-word RAM to store the PED value sorted in the sorter 220. Here, since the values output from the sorter 220 are differently output according to four modulation schemes, the alignment controller 230 may not use memory logic in memory A and B 240 and 250 according to each modulation scheme. In this case, the power consumption can be reduced by operating with a clock gating technique that stops operation. For example, in 16-QAM and 64-QAM modes, the values sorted by the sorter 220 are stored in the memories A and B 240 and 250 which are 4-word RAMs. On the other hand, in the BPSK and QPSK modes, the memory A and B (240 and 250) operate in the clock gating scheme, and in the 16-QAM mode, the memory B 250 operates in the same manner in the clock gating technique.

Meanwhile, the structure of the aligner 220 having 16 inputs and supporting four modulation schemes of BPSK, QPSK, 16-QAM, and 64-QAM is illustrated in FIG. 6.

FIG. 6 is a control block diagram of an aligner according to an embodiment of the present invention, illustrating a configuration that may be implemented in the aligner having 16 inputs of FIG. 5.

As shown in FIG. 6, the sorter 220 of the present invention includes a buffer 222 in which an input signal is buffered, and a plurality of comparators 224 for comparing the magnitude of each input value. Sort in ascending order according to their size.

A new PED value is input to the buffer 222 through the input port p every clock cycle. The input PED values are compared in size through a plurality of comparators 224, and the PED values arranged in ascending order from p1 to p7 are output through the output port. The PED values output from the sorter 220 are stored in memories A and B 240 and 250 (see FIG. 5).

As described above, the signal detection apparatus of the present invention performs symbol detection at two levels at the time of tree search by the PED calculation unit 100, thereby reducing the number of search levels by half compared with the conventional algorithm. Performance can be improved by expanding the symbol search space to find potential candidates. In addition, the alignment unit 200 operates the internal memory using a clock gating method according to the modulation method of the input signal to align the PED values output from the PED calculator and output K _i symbol pairs { s _j , s _{j -1} }. Thus, the size can be minimized while supporting various modulation schemes.

Although the present invention has been described in detail with reference to exemplary embodiments above, those skilled in the art to which the present invention pertains can make various modifications to the above-described embodiments without departing from the scope of the present invention. Will understand. Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined not only by the claims below but also by the equivalents of the claims.

1 is a node selection state diagram according to a conventional sphere decoding method

2 is a control block diagram of a signal detection apparatus of a multiple input multiple output system according to an embodiment of the present invention;

3 is a control block diagram of the PED calculation unit of FIG.

4 is an optimal candidate node selection state diagram of the PED calculation unit of the present invention;

5 is a control block diagram of the alignment unit of FIG. 1;

6 is a control block diagram of the aligner of FIG.

*** Explanation of symbols for the main parts of the drawing ***

 6: signal generator 10: controller

18: output buffer 100: PED calculation unit

200: alignment unit 210: multiplexer

220: sorter 222: buffer

224: comparator 230: alignment control unit

240: memory section A 250: memory section B

Claims (6)

An input memory unit for storing an input signal; A signal generator for generating a signal for calculating a signal for a partial Euclidean distance (PED) operation; A PED calculator for calculating a PED value based on a symbol of a node retrieved at a predetermined level of each node and a symbol of a node retrieved at a previous level according to a signal input from the input memory unit and the signal generator; An alignment unit for selecting K PEDs calculated in the PED operation unit in order of decreasing size and searching for a corresponding pair of symbols; A symbol pair storage unit for storing the symbol pairs found in the alignment unit, When the PED calculation unit is at level i = M, M-1, M-2, ..., 1. and PED L _i (s ^(j-1) ) at the level i , Equation 6 And a PED value is calculated based on a symbol of a node retrieved at a previous level of a tree search and a symbol of a node retrieved at a current level using the PED. &Quot; (6) "

Where j = 2i,

Is the partial vector of symbols found at the previous level of tree traversal. delete An input memory unit for storing an input signal; A signal generator for generating a signal for calculating a signal for a partial Euclidean distance (PED) operation; A PED calculator for calculating a PED value based on a symbol of a node retrieved at a predetermined level of each node and a symbol of a node retrieved at a previous level according to a signal input from the input memory unit and the signal generator; An alignment unit for selecting K PEDs calculated in the PED operation unit in order of decreasing size and searching for a corresponding pair of symbols; A symbol pair storage unit for storing the symbol pairs found in the alignment unit, When the PED calculation unit is at level i = M, M-1, M-2, ..., 1. and PED L _i (s ^(j-1) ) at the level i , Equation 7 And a PED value is calculated based on a symbol of a node retrieved at a previous level of a tree search and a symbol of a node retrieved at a current level using the PED. [Equation 7]

here,

Section

The term is only affected by s ^{(j + 1)} detected at the previous level. An input memory unit for storing an input signal; A signal generator for generating a signal for calculating a signal for a partial Euclidean distance (PED) operation; A PED calculator for calculating a PED value based on a symbol of a node retrieved at a predetermined level of each node and a symbol of a node retrieved at a previous level according to a signal input from the input memory unit and the signal generator; An alignment unit for selecting K PEDs calculated in the PED operation unit in order of decreasing size and searching for a corresponding pair of symbols; A symbol pair storage unit for storing the symbol pairs found in the alignment unit, The alignment unit may include a multiplexer for selectively outputting PED values output from the PED calculator; A sorter for sorting and outputting the PED values inputted through the multiplexer in ascending order; Memory A and memory B for storing PED values output from the sorter; According to a modulation method of the input signal, the memory A and the memory B includes an alignment control unit for controlling to operate by a clock gating technique that stops operation when the memory logic is not used, multiple input multiple output Signal detection device of the system. The method of claim 4, wherein The sorter, A buffer which receives a PED value from the multiplexer every predetermined clock cycle; And a plurality of comparators for comparing the PED values inputted to the buffer and outputting the PED values in an ascending order. The method of claim 4, wherein The signal detection apparatus of the multiple input multiple output system, characterized in that the modulation method of the input signal comprises at least one of BPSK, QPSK, 16-QAM, 64-QAM.

Images