KR101498267B1 - Sphere decoding method and system for signal reconstruction in mimo system - Google Patents

Abstract

A sphere decoding method for signal reconstruction in a multiple input multiple output (MIMO) system according to the present invention comprises: a reception unit receiving at least a reception symbol; a tree construction unit constructing a tree by utilizing constellation information and antenna information; and a detection unit detecting a candidate node corresponding to the reception symbol as independent per each level of the tree. According to the present invention, the complexity of sphere decoding can be remarkably reduced regardless of any constellation. In addition, the present invention can be commonly applied to all modulation methods as independent of a structure and characteristics of a symbol vector. Furthermore, advantages of increasing memory use of a receiver and decreasing the complexity of signal processing can be expected since only a single sphere decoding technique is included in the receiver.

Description

TECHNICAL FIELD [0001] The present invention relates to a method and a system for decoding a signal of a mimo system,

The present invention relates to a mimo system, in other words a sphere decoding method and system for recovering signals of a multi-antenna system.

BACKGROUND OF THE INVENTION Wireless terminals are rapidly becoming popular in recent years, and the amount of data passing through wireless terminals is increasing explosively. Of these data, high-quality video and sound are becoming important factors for increasing the communication capacity.

Speed wireless communication service is requested under the background, there is no redundancy in the wireless communication frequency currently in use, and it is also difficult to secure a bandwidth of radio frequency. As a way to solve this problem, the Multiple Input Multiple Output (MIMO) technology is once again in the spotlight.

The mimo technology can increase the communication capacity by increasing the number of antennas at the transmitter and receiver without additional bandwidth allocation. However, as the number of antennas of the transmitter and the receiver increases, the complexity of the signal processing to be processed by the receiver in order to detect a transmission symbol transmitted from the transmitter in the receiver symbol received from the receiver is exponentially .

A ML (Maximum Likelihood) detection technique has been proposed as a method of detecting a transmission symbol from a conventional reception symbol. The ML detection technique is a technique for calculating and comparing Euclidean distances with respect to all transmittable symbol vectors corresponding to a modulation scheme in order to detect transmitted symbols, and is an excellent technique in terms of performance. However, as the number of antennas and the modulation method increase, the complexity increases exponentially, which makes the implementation very difficult. A sphere decoding technique has been developed to reduce the complexity of ML detection.

The Sphere decoding scheme reduces the complexity of ML detection by calculating the Euclidean distance only for a symbol vector set existing in a sphere having an initially set radius in consideration of noise variance and channel state. However, there is a problem that the complexity varies depending on the initial radius of the sphere decoder. In other words, if the initial radius is set too large, a large number of lattice vectors exist within the initial radius and have almost the same complexity as the ML detector. Also, if the initial radius is too small, a valid grid vector can not be found. In addition, the lower the SNR of the sphere decoder, the more the number of visited nodes, i.e., the complexity, increases at the time of tree search, and the decoding efficiency deteriorates. Due to these factors, the complexity of signal processing is still high. This problem is becoming a big problem as the number of antennas increases in the beauty technology.

Real-Valued Sphere Decoding (SAD) technology has been proposed as a method for reducing the complexity of the Sphere decoding technique. The real-SD technique is a technique for simplifying the complexity by separately processing the imaginary part and the real part of the symbol vector. However, since the real-SD method has a limitation on the constellation according to the modulation method, there is a problem that can be applied only to the rectangular QAM constellation. Likewise, even in complex-SD (Complex Valued SD), there is a problem that can be applied only to a square QAM and a PSK constellation. In addition, both the real number SD and the complex number -SD increase in complexity as the constellation becomes complicated, and there is a problem that it can be applied to only one constellation or one constellation.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a mimo system capable of reducing the complexity of signal processing and applicable to all constellations and processing all constellations with a single technique without changing the decoding method. And a system therefor are proposed.

A system for decoding a signal of a mimic system according to the present invention includes: a receiver for receiving at least a received symbol; A tree component for constructing a tree using constellation information and antenna information; And a detector for detecting a candidate node corresponding to the received symbol independently of each level of the tree.

이고, 여기서,

이고,

이고, x는 원형제약의 중심점(X는

r, r은 받은 수신신호)이고, s는 송신심벌이고, H는 채널정보이고, k는 안테나의 레벨(K는 1~n 사이의 값이고, Sk는 S벡터의 k번째요소)인 것을 특징으로 한다. 이로써 정확한 레벨별 제약조건을 부여할 수 있다. In the decoding system, the candidate node is selected by a circular constraint,

Lt; / RTI >

ego,

, X is the center point of the circular constraint (X is

r is the received signal), s is the transmitted symbol, H is the channel information, k is the antenna level (K is a value between 1 and n, Sk is the kth element of the S vector) . This gives the exact level-specific constraints.

In the decoding system, the detection unit detects a candidate node in a range satisfying a set value by using a two-dimensional circular constraint. This makes it possible to distinguish candidate nodes with simple calculations. Meanwhile, the detection unit further divides the candidate constellations for each level, and then constructs candidates of the n-dimensional vector signal considering all the level candidates. Then, the n-dimensional vector signal candidates are multiplied by the H matrix, and the distance between the values and the received signal is calculated. Then, the smallest distance among them can be predicted by the transmission signal. In this way, symbol vectors for all levels can be determined.

According to another aspect of the present invention, there is provided a sphere decoding system for signal restoration of a mimic system, which detects a candidate node independently of each level of a tree, multiplies a symbol of the lowest node in the tree by an H matrix, And a detector for selecting the shortest one.

According to another aspect of the present invention, there is provided a method of decoding a signal of a mimic system of the present invention, comprising: receiving at least a received symbol, constellation information, and antenna information; Constructing a tree using the constellation information and the antenna information; And detecting candidate nodes independently for each level of the tree.

In the decoding method, the candidate node satisfies a set value using the two-dimensional constraint at each level. This makes it possible to quickly search candidate nodes.

According to the present invention, even if the number of antennas is increased, the complexity does not increase exponentially. Therefore, the complexity of the signal processing in the receiver is reduced and the signal processing can be performed on all constellations with a single decoding technique without any difference according to the constellation The advantage can be expected.

1 is a configuration diagram of a mimo system according to an embodiment;
2 is a configuration diagram of a sphere decoder according to an embodiment;
3 is a tree used for Sphere coding according to an embodiment;
4 is a flow chart illustrating a sphere decoding method for signal recovery of a mimic system according to an embodiment;

Hereinafter, specific embodiments of the present invention will be described in detail with reference to the drawings. It should be understood, however, that the scope of the present invention is not limited to the embodiments described below, and other embodiments may easily be suggested by adding, changing, deleting, and adding elements, I would say.

1 is a configuration diagram of a mimo system according to an embodiment.

Referring to FIG. 1, a fine system according to an exemplary embodiment includes a transmitter 1 for transmitting independent transmission symbols for each of at least two transmission antennas, and a transmitter 1 for detecting transmission symbols from received symbols for each of at least two reception antennas. And a receiver (2) for performing the same.

The transmitter 1 encodes and interleaves a bit string corresponding to an input signal to be transmitted, and then performs serial-to-parallel conversion according to the number of antennas. The parallel-converted lattice-form transmission symbols are simultaneously transmitted through the respective antennas. The receiver 2 receives a transmission symbol transmitted from a plurality of antennas included in the transmitter 1, detects a plurality of independent transmission symbols included in the received symbol, and outputs a detection signal. It will be appreciated that the received symbols differ from the transmitted symbols according to the communication environment between the transmitter 1 and the receiver 2. [ The receiver 20 includes a sphere decoder 3 for detecting transmitted symbols.

2 is a configuration diagram of a sphere decoder according to an embodiment.

Referring to FIG. 2, the sphere decoder 3 includes a receiver 31 for receiving constellations according to a plurality of reception symbols and a modulation method, a tree for spear decoding using antenna related information and constellation related information A tree structure unit 32, and a detection unit 33 for detecting transmitted symbols.

A spear decoding method implemented through modeling of a mimosa system will be described.

The channel model of the mimo system when the number of antennas of the receiver 2 is M and the number of antennas of the transmitter 1 is N can be expressed as follows.

은 수신된 신호를 나타낸다. H는 M×N 블록 레일리 페이딩 채널 행렬(Rayleigh fading channel matrix)를 나타낸다. H의 각 엔트리들은 IID(independently and identically distributed) 복소수 제어-평균 유닛 분산 가우시안 랜덤 변수이다. 여기에서 채널 행렬은 수신기(2)에 알려져 있는 값일 수 있다.

는 송신기(1)로부터 송신된 심벌벡터를 나타내며,

을 만족한다. 여기에서

는 복소수 성좌를 지칭하며,

는 복소수 도메인을 나타낸다.

은 AWGN(additive white Gaussian noise)벡터를 나타내며, 제로-평균, 분산

을 가진다.From here,

Represents the received signal. H denotes an M × N block Rayleigh fading channel matrix. Each entry of H is an independently and identically distributed (IID) complex control-mean unit variance Gaussian random variable. Here, the channel matrix may be a value known to the receiver 2.

Denotes a symbol vector transmitted from the transmitter 1,

. From here

Refers to a complex constellation,

Represents a complex number domain.

Represents an AWGN (additive white Gaussian noise) vector, and the zero-mean, variance

.

는 행렬 H의 k번째 열(row)을 나타내고,

는 행렬 H의 k번째 행(column)을 나타내고,

는 행렬 H의 복합공액전치행렬을 나타내고,

는 H의 수도(pseudo)역수를 나타낸다. 벡터의 각 성분들은 첨자로 표시된다. 예를 들어

는 S의 k번째 성분을 나타낸다. 한편,

은 s의 마지막 N-k+1요소를 취하는 벡터이고,

은 집합

의 카디널리티(cardinality)이고,

은 벡터 S의 2nd 놈(norm)이다.

Represents the kth column of the matrix H,

Represents the kth column of the matrix H,

Represents a complex conjugate transpose matrix of matrix H,

Represents the pseudo inverse of H. Each component of the vector is represented by a subscript. E.g

Represents the kth component of S. Meanwhile,

Is a vector that takes the last N-k + 1 elements of s,

Is a set

Of cardinality,

Is the second norm of the vector S.

In the mimo system, the receiver 31 of FIG. 2 receives constellation-related information and a reception symbol. Thereafter, the tree structure unit 32 constructs a tree by referring to the number of antennas and constellation related information. The tree structure unit 32 may form a tree by any method or may construct a tree in the order that is likely to be pruned.

A general Sphere decoding method can be performed by 1) identifying a symbol vector satisfying a predetermined constraint as a candidate, and 2) selecting a candidate having a minimum distance from the received signal among candidate symbol vectors.

As the above constraint, a spherical constraint (SC) can be used. The spherical constraint can be expressed as Equation (1).

이다. here,

to be.

Equation (1) shows that a symbol vector satisfying this constraint is selected by comparing the distance d (s) to all the symbol vectors with the set value C. However, calculating the Euclidean distance d (s) in the above equation (1) is not preferable because two calculations are required. Therefore, Equation (1) can be replaced with Equation (2).

은

의 QR 분해(decomposition))에서 어퍼 삼각 행렬(upper triangular matrix)이고,

이고,

일때

이고, x는 중심점이다. 그러므로, k>l일때, 거리계산는

값이 변할 때 중복해서 수행할 필요가 없다. 따라서,

는 부분 유클리드 거리(parital euclidean distance:PED)라고 한다. In Equation (2)

silver

The upper triangular matrix in the decomposition of QR,

ego,

when

And x is the center point. Therefore, when k> l, The

There is no need to duplicate when the value changes. therefore,

Is called the parical euclidean distance (PED).

Equation (2), which does not require additional calculation, can be used to determine whether or not the spherical constraint of the sphere decoding is satisfied, instead of Equation (1). A node on the tree that does not satisfy the spherical constraint can significantly reduce the complexity by performing pruning. At this time, for a node that has been pruned, search may not be performed for a node smaller than the pruning node, thereby reducing the complexity.

However, even in the case of Equation (2), the complexity is still high, and the distance calculation of the k-th level is also dependent on the elements of S other than Sk, and is dependent on the level of the tree. In order to further reduce these problems, attempts have been made to simplify the old constraints (SC), and the simplification of the old constraints has been based on utilizing the features and structures of specific constellations. As a result, it has evolved to propose a suitable spear decoding technique only for a specific constellation. Conversely, as already mentioned, certain sphere decoding techniques have been applied only to modulation schemes having a particular constellation, and can not be applied to other constellations. Therefore, there is a problem in that a decoding technique suitable for a constellation must be separately recorded in the receiving end. Also, there is a problem of occupying the memory capacity of the receiving end. There is also a problem of consuming additional time for calling and signal processing of individual decoding methods.

In the embodiment, a two-dimensional circular constraint (CC) condition that is not constrictive to the constellation and is applicable to any constellation and whose complexity is remarkably reduced and which is not dependent on the level of the tree is proposed. First, the condition of Equation (1) can be replaced with the following Equation (3).

Equation (3) can be transformed as shown in Equation (4). Here, it has been described that x is a virtual center of a spherical constraint.

가

에서 풀랭크를 가질 때

가 성립한다는 사실이 적용된 결과이다. In Equation (4), (a) is the application result of the Kosi-Schwartz inequality, (b)

end

When you have a full rank in

Is the result of applying the fact that.

Equation (3) and Equation (4) are combined and the results are summarized as Equation (5).

이고,

이다. 여기서, x는 원형제약의 중심점(X는

r, r은 받은 수신신호)이고, s는 송신심벌이고, H는 채널정보이고, k는 안테나의 레벨(K는 1~n 사이의 값이고, Sk는 S벡터의 k번째요소)이다. 상기 수학식 5의 조건을 원형제약(circular constraint)라고 할 수 있고, 상기 원형제약을 만족하는 심벌벡터가 상기 트리구성부(32)에 구성된 트리의 각 레벨의 후보노드로서 선택될 수 있다. here,

ego,

to be. Where x is the center point of the circular constraint (X is

r is the received signal), s is the transmitted symbol, H is the channel information, k is the level of the antenna (K is a value between 1 and n, Sk is the kth element of the S vector). The condition of Equation (5) may be referred to as a circular constraint, and a symbol vector satisfying the circular constraint may be selected as a candidate node at each level of the tree constructed in the tree structure unit 32. [

는 스피어 디코딩의 검색이 수행되기 전에 계산될 수 있을 뿐만 아니라,

행렬이 변하기 전까지는 변하지 않는다. 그러므로 채널정보가 동일한 때에는 동일한 값이 사용될 수 있다. 상기

는 C-메트릭(metric)이라고 하기로 한다. 상기

는 현재 레벨(k)(안테나의 레벨일 수 있다)에만 의존하는 값으로서 다른 레벨(l:

)에는 영향을 받지 않는다. 상기 x는 수신심벌과 관련되는 정보로서 원형제약의 중심점이 될 수 있다. 그러므로, C-메트릭을 계산함에 있어서 다섯번의 부동소수점연산(FLOPs)만이 필요하기 때문에, 10(N-k)+8번의 부동소수점연산이 필요한 PED 계산에 비하여 계산량이 줄어든다. 또한, 트리 상에서 현재 레벨의 모든 노드에 대해서만 원형제약을 적용하므로 c-메트릭의 계산량이 줄어든다. remind

Not only can the calculation of the sphere decoding be performed before calculation,

It does not change until the matrix is changed. Therefore, the same value can be used when the channel information is the same. remind

Is called a C-metric. remind

Is a value dependent only on the current level k (which may be the level of the antenna)

) Are not affected. X can be the center point of the circular constraint as information related to the received symbol. Therefore, since only five floating point operations (FLOPs) are required in calculating the C-metric, the computational complexity is reduced compared to the PED computation requiring 10 (Nk) + 8 floating point operations. In addition, the circular constraint is applied only to all the nodes of the current level in the tree, thereby reducing the calculation amount of the c-metric.

Because of the advantages described above, circularity constraints can be expected to greatly reduce the complexity. The detection unit 33 of FIG. 2 performs the search by applying the circular constraint and obtains a symbol vector for each level on the tree. Hereinafter, the application of the circular constraint will be described in detail using a tree as an example.

FIG. 3 shows an example of a tree used for spear coding. FIG. 3 shows an example of a constellation having four transmitting antennas and two symbol vectors (for example, a BPSK modulation scheme). Of course, other mimo systems and other modulation schemes are equally applicable.

At each level of the tree, nodes satisfying the circular constraints are indicated. Specifically, when the level k is 1, the third node from the left side of the tree satisfies the circular constraint. When the level k is 2 and 3, the left two nodes satisfy the circular constraint, (k) is 4, it can be seen that all nodes meet the circular constraint. In such a case, it can be confirmed that the distance of the bold solid line in FIG. 3 is the shortest, and this information can be determined as the transmission symbol.

If the node satisfying the circular constraint at each level is excessively large or small, the set value C of Equation (5) can be made large or small. This allows an appropriate number of nodes to be derived. As another example, when there are at least two cases in which nodes satisfying a circular constraint at each level are connected to each other on a level basis, the symbol of the lowest node in the tree is multiplied by H, You can choose. In other words, the candidate constellations are separated for each level, and then the candidates of the n-dimensional vector signal are considered in which all the level candidates are considered. Then, the n-dimensional vector signal candidates are multiplied by the H matrix, and the distance between the values and the received signal is calculated. The smallest distance among them can be predicted by the transmission signal.

In any case, since the node satisfying the circular constraint for each level can be separately obtained in the embodiment, the complexity can be remarkably reduced. Further, since it is not dependent on the structure or characteristic of the symbol vector, it can be applied to all modulation schemes. In this case, the modulation scheme is known to the receiver. In addition, since only a single technique is stored in the receiving end, it is expected that the memory utilization of the receiving end is increased and the complexity of signal processing is reduced.

4 is a flowchart illustrating a sphere decoding method for signal restoration of a mimic system according to an embodiment. Referring to FIG. 4, in the mimo system, at least information related to at least two reception antennas, constellation, and transmitter antennas is received from at least two antennas (S1). A tree is constructed with reference to the antennas of the constellation and the transmitting end (S2), and the candidate constellations are separated for each level, and candidates of the n-dimensional vector signal are then formed considering all the level candidates. Then, the n-dimensional vector signal candidates are multiplied by the H matrix, and the distance between the received signals and the received signal is calculated.

In searching for the node, a two-dimensional circular constraint (CC) shown in Equation (5) is utilized. A symbol vector satisfying the circular constraint can be selected as a candidate node of each level of the tree.

If the node satisfying the circular constraint at each level is excessively large or small, the set value C of Equation (5) can be made large or small. This allows a reasonable number of nodes to be retrieved. In another case, when there are at least two cases in which nodes satisfying the circular constraint at each level are connected to each other, the symbol having the lowest distance in the received symbol can be selected after multiplying H by the symbol of the lowest node in the tree.

According to the present invention, the complexity of sphere decoding can be significantly reduced in the case of any constellation. Also, since it is not dependent on the structure or characteristic of the symbol vector, it can be commonly applied to all modulation schemes. In addition, since only a single sphere decoding technique is recorded in the receiving end, the memory utilization of the receiving end is increased and the complexity of the signal processing is reduced.

Therefore, even if the number of antennas is increased, it can be sufficiently coped with, and it is demanded to be urgently applied to a high-speed wireless communication network.

3: Sphere decoder

Claims (7)

In the mimo system,
A receiving unit for receiving at least a received symbol;
A tree component for constructing a tree using constellation information and antenna information; And
And a detector for detecting a candidate node corresponding to the received symbol with a circular constraint independently for each level of the tree. The method according to claim 1,
The candidate node is selected by the circular constraint,
The circular constraint,

ego,
here,

ego,

, X is the center point of the circular constraint (X is

r is the received signal), s is the transmitted symbol, H is the channel information, k is the level of the antenna (K is a value between 1 and n, Sk is the kth element of the S vector)
Spear decoding system for signal recovery of mimo system. The method according to claim 1,
Wherein the detection unit detects a candidate node in a range satisfying a set value by using a two-dimensional circular constraint. The method of claim 3,
Dimensional vector signal candidates considering all of the candidate constellations, multiplying the n-dimensional vector signal candidates by an H matrix, and comparing distances between the constellation values and the received signal, And estimating the smallest one as a transmission signal. In the mimo system,
The candidate nodes are detected through circular constraints independently for each level of the tree,
And a detection unit for multiplying a symbol of the lowest node in the tree by an H matrix and selecting a symbol having the shortest distance from the received symbol. Receiving at least a received symbol, constellation information, and antenna information;
Constructing a tree using the constellation information and the antenna information; And
And detecting a candidate node through a circular constraint independently for each level of the tree. The method according to claim 6,
Wherein the candidate node satisfies a set value using two-dimensional constraints at each of the levels.

Images