KR102074821B1 - Low-complexity Sphere Search-based Adaptive Spatial Modulation method and MIMO system using the same - Google Patents

Abstract

The present invention relates to a low complexity adaptive spatial modulation method using spherical search as a spatial modulation technique of a multi-antenna system.

Description

Low-complexity Sphere Search-based Adaptive Spatial Modulation method and MIMO system using the same

The present invention belongs to the field of spatial modulation technology of multiple antenna systems.

Spatial modulation (SM) is an energy efficient signal transfer technology that handles the tradeoff between multiplexing and diversity gain in a multiple-input multiple-output (MIMO) system.

The SM technique achieves high overall spectral efficiency (SE) by using the antenna index dimension of the MIMO system, and activation at each instant of the unit transmit antenna significantly increases hardware complexity by requiring only a single RF chain at the transmitter. To reduce. The SM technique also solves internal channel interference and internal antenna synchronization problems.

In order to improve the performance of the SM-MIMO system, an adaptive spatial modulation (ASM) technique has been proposed that combines SM and adaptive modulation to improve SE or improve bit error rate (BER). Aim to maximize.

Another ASM technique called modulation order selection (MOS) has been proposed to achieve constant data rates and to improve BER performance.

Specifically, to maximize the free distance (FD) that leads to the minimum BER, the MOS algorithm performs an exhaustive search (ES) on the ASM set and all the error vectors of each ASM.

To determine the near optimal solution at low complexity, a simplified ASM technique called candidate reduction (CR) and one-bit reallocation (OBRA) has been proposed.

This limits the search space in ASM mode to a smaller set based on ASM mode and their likelihood of variation. The numerical analysis shows that only a small degradation in BER performance occurs when compared to an optimized technique based on small MIMO systems.

As described above, the spatial modulation (SM) is a technique of transmitting a signal using only one antenna among transmission antennas of a multi-antenna system and transmitting information using an index of an antenna used for transmission together with the transmitted symbol. By adaptively selecting the modulation scheme for each antenna according to the channel state (adaptive spatial modulation), the reception error rate performance can be improved.

However, the conventional method has a problem in that a high calculation amount is required to select an optimal modulation method.

Korean Patent No. 1771503 includes the steps of: a receiving sequencing unit aligning indices of receiving antennas in order of an expected value of Euclidean distance calculated for a signal received by each receiving antenna; The transmission ordering unit aligning the indices of the transmitting antennas in the order of the output of the maximum ratio combining filter; And a receiving space searching unit calculating a transmitting antenna and a symbol by performing a receiving space search, wherein the Euclidean distance value is a geometric distance between two points in an n-dimensional space, and is an average of the receiving antenna. Spatial modulation is performed based on the symbol output, the received signal value, the transmission vector to the receiving antenna, and the number of transmitting antennas.

As such, among the various problems of the generalized spatial modulation technology, there is a signal processing technique that presents a problem of reducing the amount of computation or calculation of the system. However, some methods using sphere search (SS) are used as a means of solving the signal processing. No scheme has been proposed to optimize computational complexity by searching only error vectors.

The present invention proposes a low-complexity adaptive spatial modulation method using spherical search to reduce the computational complexity of a technique for finding an optimal modulation scheme for adaptive spatial modulation.

The present invention provides an adaptive spatial modulation (ASM) method for adaptively selecting a modulation scheme for each transmitting antenna according to a channel state in a multi-antenna system. When searching for an optimal modulation scheme for adaptive spatial modulation, the search space is spherical. An Old Search Adaptive Spatial Modulation (SS-ASM) method is characterized by searching only a part of an error vector by using a sphere search (SS) which is reduced to an error vector.

According to an aspect of the present invention, the old search is applied by modifying a sphere decoding receiver for generalized spatial modulation.

According to an aspect of the present invention, the old search applies a search early termination method by setting a threshold.

According to another aspect of the present invention, there is provided a multiple input and output (MIMO) system applying the old search adaptive spatial modulation (SS-ASM) method.

According to the present invention, the complexity of the calculation is lowered by 90% or more by narrowing the error vector search area by using the old search without searching all the error vectors in order to find the optimal modulation scheme, thereby having high practicality.

It is applicable to next-generation communication systems such as low-power IoT technology, which is a communication technology that has the advantages of low complexity and low power of spatial modulation by increasing spatial modulation performance and reducing complexity through adaptive modulation.

1 is an algorithm of an old search adaptive spatial modulation (SS-ASM) method according to the present invention.

In order to reduce the computational complexity while maintaining the system performance of the ASM algorithm, the present invention proposes an ASM (SS-ASM) technique using old search, which effectively reduces the search space of the error vector of each ASM mode.

This is because in an ASM where each error vector contains only up to two non-zero inputs, the search process of the error vector is similar to symbol search in generalized spatial modulation (GSM).

Based on this fact, it is possible to apply an SS process similar to that of a decoded receiver for GSM systems to search for error vectors and to efficiently calculate free distance. In addition, the present invention proposes a threshold to terminate the SS process early to further reduce the amount of computation.

Hereinafter, an adaptive spatial modulation (SS-ASM) method using sphere search according to the present invention will be described in detail.

의 최대값을 가진 ASM 모드를 결정하기 위해, 최적의 ASM 알고리즘은 모든 ASM모드를 검사하고, 각 ASM 모드에 대해 모든 가능한 전송 신호 벡터를 위한 ES를 수행하는바, 이는 상당한 계산량을 요구한다.FD Inn

To determine the ASM mode with the maximum of, the optimal ASM algorithm examines all ASM modes and performs ES for every possible transmission signal vector for each ASM mode, which requires considerable computation.

Accordingly, the present invention provides an error vector in a sphere to reduce computational complexity in an adaptive spatial modulation (ASM) method that adaptively selects a modulation scheme for each transmit antenna according to channel conditions in a multi-antenna system. We efficiently calculate the FD for each ASM mode using the Sphere Search (SS) technique, which reduces and searches only some error vectors.

 Before describing the SS-ASM algorithm according to the present invention, the SD for the GSM system will be described.

The SD-GSM technique was originally designed to efficiently search for symbols in GSM systems, but it will be shown that the present invention can be applied to search for FD for ASM with some modification.

1.SD-GSM

개의 송신 안테나와

개의 수신안테나를 가진 MIMO 시스템을 전제로 한다.For GSM, similar to the system model for ASM,

Transmission antennas

It is assumed that a MIMO system having two receiving antennas.

개의 송신 안테나 중

가 동시에 구동되고, 그리고 나서, 송신 신호

는

개의 비 제로 요소를 포함한다. 그리고 수신 신호는 아래 식으로 주어진다.In GSM, however, multiple transmit antennas are driven at each instant. Specifically, the present invention

Of transmitting antennas

Are driven simultaneously, and then the transmission signal

Is

Non-zero elements. And the received signal is given by

<Equation 1>

×

채널 행렬을 나타내며, 각 요소들은 영평균과 단위 변동량을 가진 독립적이고 동일하게 분산된 복합 가우시안 랜덤 변수로 가정한다.Where H is

×

Represents a channel matrix, each of which is assumed to be an independent and equally distributed complex Gaussian random variable with zero mean and unit variance.

를 가진 영평균 복합 가우시안 입력으로 구성된 노이즈 벡터이다.n is the variation per dimension

A noise vector consisting of a zero-average complex Gaussian input with.

는 다음 식2와 같다.Optimal likelihood (ML) solution

Is shown in Equation 2 below.

<Equation 2>

는 GSM시스템에서의 모든 가능한 송신 신호벡터 집합이다. ML 해는 다음 식3과 같이도 나타낼 수 있다.here,

Is the set of all possible transmit signal vectors in the GSM system. The ML solution can also be expressed as

<Equation 3>

는 영강압 추정치를 나타낸다. Cholesky 분해를 통해, 상 삼각의

×

행렬인 U를 얻을 수 있고, 이는

를 만족한다.here,

Represents the zero crush estimate. Through Cholesky decomposition, the upper triangular

×

We can get the matrix U, which is

Satisfies.

Therefore, Equation 3 can be expressed as Equation 4 below.

<Equation 4>

To reduce the computational complexity to find the ML solution, the SD receiver examines only the vectors in the sphere of radius d. That is, the present invention only needs to check the signal vector satisfying the constraint of Equation 5 below.

(Eq. 5)

If we expand the above equation, we can express it as

<Equation 6>

는 s의 i번째 입력이고,

는 U의 i번째 행 j번째 열 입력이다.here,

Is the i th input of s,

Is the j-th column input of the i-th row of U.

에서 시작하고, 레벨 i에서 식7의 조건을 검사한다.In SD, a depth first tree search is performed to find a symbol vector that satisfies the constraint of Equation 5. Ie SD level

Start at, and check the condition of equation 7 at level i.

<Eq. 7>

이며,

는 레벨 i에서의 구 반지름을 나타내고, 이는 다음 식8에 의해 반복적으로 계산된다.here,

Is,

Represents the radius of the sphere at level i, which is calculated iteratively by

(Eq. 8)

는 d로 설정된다. 상기 식7의 조건이 만족되면, 탐색 프로세스는 레벨 1에 도달할 때까지 레벨 i-1로 진행한다.here,

Is set to d. If the condition of Equation 7 is satisfied, the search process proceeds to level i-1 until level 1 is reached.

 In contrast to SD for spatial multiplexing, the SD-GSM technique must consider the possibility of non-transmission from each transmit antenna during the tree search process.

레벨로부터의 레벨(

)은

= 0이라는 심볼 값을 갖고, 남아있는

레벨은 비 제로 신호

을 갖는다. 반면

은 컨스텔레이션 집합을 가리킨다. 가능한 심볼 후보를 추적하기 위해, GSM-SD기법은 레벨i에서의 조합 행렬(식9)을 적용한다.Especially full

Level from level (

)silver

Has a symbol value = 0 and remains

Level is non-zero signal

Has On the other hand

Indicates a constellation set. To track possible symbol candidates, the GSM-SD technique applies a combination matrix at level i (Equation 9).

<Eq. 9>

개의 가능한 활성 안테나 조합을 포함하고, 여기서

이진벡터인

는 유효한 안테나 조합을 나타낸다.this is

Two possible active antenna combinations, wherein

Binary vector

Denotes a valid antenna combination.

인 동안,

는 다음 식10과 같이 4개의 가능한 안테나 조합으로 구성된다.For example,

While being

Is composed of four possible antenna combinations as

<Eq. 10>

을 가리키고, 0은

= 0을 가리킨다.Where in the i row, 1 is

, 0 is

= 0.

로 갱신된다. 위 예에서

가 선택된다면, 본 발명은 네 번째 입력을 위해 0을 갖는 열과 같이 무효한 안테나 조합을 배제함으로써

을 얻을 수 있다.After selecting the symbol candidate at level i, the collation matrix is

Is updated to In the example above

If is selected, the invention eliminates invalid antenna combinations, such as columns with zeros for the fourth input.

Can be obtained.

은 다음 식11과 같이 표현된다.As a result,

Is expressed by Equation 11 below.

<Eq. 11>

= 0이 선택된다면,

이 아래 식12로 설정된다.In contrast,

If = 0 is selected,

This is set by Equation 12 below.

<Eq. 12>

= 0이 레벨 3에서 선택되는 경우,

를 얻을 수 있고, 이는 레벨 1과 레벨 2에서 오직

의 심볼만이 유효함을 의미한다. 그러므로

= 0 과

= 0은 탐색 지역에서 배제된다.If we continue with Eq. 12,

If = 0 is chosen at level 3,

Is obtained, which is only available at level 1 and level 2

Means that only the symbol is valid. therefore

= 0 and

= 0 is excluded from the search area.

2. SS-ASM

FD of the k-th ASM mode may be expressed by the following equation (13).

<Eq. 13>

는 k번째 ASM 모드를 위한 모든 가능한 송신 신호 벡터의 집합이다.here,

Is the set of all possible transmit signal vectors for the kth ASM mode.

로 설정될 때 식13이 식2와 같아짐을 나타낸다.The comparison between Equation 13 and Equation 2 is

Equation 13 is equal to Equation 2 when set to.

That is, the process of finding the FD in Equation 13 may be converted to the ML symbol search process in Equation 14.

<Equation 14>

는 k번째 ASM 모드를 위한 모든 가능한 오류 벡터

이고,

이다. 오류 벡터는 오직 두 개의 비 제로 입력만을 갖고 있는데 xi 와 xj는 단일 비 제로 입력을 가지며, 이는 오류 벡터가 GSM의 송신 신호에서와 유사한 구조를 갖고 있기 때문이다.here,

Is all possible error vectors for the kth ASM mode

ego,

to be. The error vector has only two nonzero inputs, because xi and xj have a single nonzero input because the error vector has a structure similar to that of the GSM transmit signal.

Therefore, in the present invention, by applying the SS process developed by modifying the above-described SD-GSM to modify the sphere decoding receiver for generalized spatial modulation (GSM) for the ASM to apply to the ASM, Equation 14 can be efficiently solved. This also means that the FD in Equation 13 can be calculated with low complexity.

By reusing or modifying Equations 5 to 8, the SS process for each ASM mode can be expressed as follows.

When applying SS to ASM, two cases need to be considered.

와

에서의 비 제로 요소가 같은 위치에 있다면, 오류 벡터

는 오직 단일 비 제로 입력을 가진다. 그러므로 식14를 풀기 위한 프로세스는

을 가진 SD-GSM 탐색기에서와 같아진다.Case 1:

Wow

If the nonzero elements at are in the same position, then the error vector

Has only a single non-zero input. Therefore, the process for solving Eq. 14

Same as in SD-GSM Explorer with.

와

에서의 비 제로 요소가 다른 위치에 있다면, s는 두 개의 비 제로 입력을 포함하고, 그로 인해 식 14를 푸는 프로세스는

를 가진 SD-GSM 탐색기에서와 같아진다.Case 2:

Wow

If the nonzero element at is in another position, s contains two nonzero inputs, so the process of solving Eq.

Same as in SD-GSM Explorer with.

를 가진 ASM시스템의 예를 든다. For Case 1 and Case 2, the combination matrix must be designed appropriately. To illustrate the combination matrix design,

Here is an example of an ASM system with:

At the initial level, the combination matrix for Case 1 can be expressed as

<Eq. 15>

On the other hand, for case 2, Equation 16 is used.

<Eq. 16>

In order to obtain SS solution considering case 1 and case 2, two possible approaches can be considered.

In the first approach, called direct search, C1 and C2 are combined to form a constructive matrix such as Eq.

<Eq. 17>

는 i = 1, 2, ···,

인 탐색프로세스 동안,

유효값을 검사하기 위해 적용된다.after,

I = 1, 2, ...

During the discovery process,

Applied to check valid values.

In contrast, in a second approach, called a two-stage search, the search is first performed within the search area corresponding to Case 1 to find the distance to the initial solution and the sphere center, denoted I R.

]은 식18에 의해 재구성될 수 있다.Note that the solution for Case 1 can be found at low complexity. Specifically, if the search space is constrained by an error vector with a single nonzero value at the lth input, then the solution of equation (14)

] Can be reconstructed by equation (18).

<Eq. 18>

은 U의 l번째 열을 나타낸다. 식18에서, 최소 진폭 오류 신호

만이 경우 1을 위한 탐색 프로세스 동안 l번째 비 제로 오류 위치로 고려될 필요가 있다.here,

Represents the lth column of U. In Equation 18, the minimum amplitude error signal

Only case 1 needs to be considered as the lth non-zero error position during the search process for 1.

개의 오류 벡터만이 경우 1의 해로서 검사되어야 하며, l번째 오류 위치를 위한 다중 최소 진폭 오류 신호가 존재할 수 있음을 유의해야 한다. As a result, considerably fewer error vectors, compared to the total number of error vectors,

Only error vectors should be checked as the solution of 1, and it should be noted that there may be multiple minimum amplitude error signals for the l th error location.

를 갖고, 그로 인해, 그들 사이의 선택은 PEP 성능에 영향을 미치지 않는다.But the error vectors are all the same distance

Therefore, the choice between them does not affect PEP performance.

After the search for case 1, I R is used as the initial radius of SS for case 2, which is performed using the combination matrix in equation 16.

In summary, cases 1 and 2 should be considered separately in the search process, and the solution for case 1 is used to reduce the initial radius of SS for case 2, which reduces the computational complexity of SS for case 2 You can.

Comparing the complexity between these two approaches, the simulation results show that the two-stage search requires less complexity than the direct search, so the complexity is analyzed based on the two-stage search.

로 표기되는 최대 FD를 가진 ASM모드는 최적 ASM 모드로 세팅되고, 이는 신호 전송을 위해 사용된다. After performing SS for ASM mode,

The ASM mode with the maximum FD denoted by is set to the optimal ASM mode, which is used for signal transmission.

In order to further reduce the computational complexity in the present invention, early termination of the SS for each ASM mode is considered.

3. Early Termination

가 대부분의 채널인식의 다른 ASM 모드에서 보다 더 크기 때문이다.Among the various ASM modes, the unified bit allocation mode is most likely to be selected, which is FD

Is larger than in most ASM modes of channel recognition.

는 초기에

으로 세팅되고,

는 더 작은 FD를 가진 ASM 모드가 발견될 때마다 갱신된다.so

Early on

Set to,

Is updated each time an ASM mode with a smaller FD is found.

와

(임계값에 해당함)를 비교한다. During the SS process for ASM mode, we check the error vector in the sphere until we find the one with the minimum distance. However, when early termination is applied during the SS process, the calculated distance of each error vector

Wow

(Corresponds to the threshold).

를 만족한다면, 상응하는 ASM의 FD는

보다 적어야 하므로, 이는 최적 ASM모드가 아님을 의미한다. As described above, the old search according to the present invention applies a search early termination method by setting a threshold value.

If it is satisfied, the corresponding ASM's FD is

Since it should be less, this means that it is not the optimal ASM mode.

을 만족한 오류 벡터가 발견되면, 현재 ASM 모드를 위한 SS 프로세스가 즉각적으로 종결되고 다음 ASM모드를 위한 작업이 시작된다.so

If an error vector that satisfies is found, the SS process for the current ASM mode is immediately terminated and work for the next ASM mode begins.

보다 더 작다면,

는 이 FD로 갱신된다.In contrast, FD in ASM mode

If smaller than

Is updated with this FD.

The overall process of SS-ASM according to the present invention can be summarized by the algorithm of FIG.

In steps 1-4, the error vectors for uba (uniform-bit-allocation) ASM mode are checked and the minimum distance between them is set at the initial FD.

In step 1, only the error vector corresponding to 1 when in uba is considered to find I R, which is used as the initial radius when performing SS for case 2 in step 2.

In steps 5-10, the ASM modes except uba are examined to determine the optimal ASM mode.

의 FD는 초기에 I R로 세팅되고, 이는 단계 6에서 발견된다. 그리고 나서 단계 13에 기재된 바와 같이, 더 작은 거리를 가진 오류 벡터가 발견될 때마다 갱신된다.In step 7,

The FD of is initially set to IR, which is found in step 6. Then, as described in step 13, it is updated each time an error vector with a smaller distance is found.

보다 작은 거리를 가진 오류 벡터가 발견될 때 수행된다.The current FD is used as the sphere radius for the SS in step 8. In step 11, early termination

This is done when an error vector with a smaller distance is found.

In steps 17 and 18, the minimum FD and optimal ASM modes are updated because new ASM modes with smaller FDs are found without early termination.

The present invention relates to a technique for selecting an optimal modulation scheme in a multiple input / output (MIMO) system, and the technical scope includes a multiple input / output (MIMO) system to which the old search adaptive spatial modulation (SS-ASM) method described above is applied.

Claims (4)

In an adaptive spatial modulation (ASM) method for searching for an optimal adaptive spatial modulation according to a channel state of a modulation scheme for each transmitting antenna in a multi-antenna system,
Searching for error vectors that are the differences of any of the transmitted signal vectors;
Set the distance of the searched error vectors as the initial free distance (FD), and then limit the sphere whose radius is the initial free distance to the search space and search only the error vector in the limited sphere. Performing a search (Sphere Search, SS); And
If the distance of the searched error vectors is smaller than the initial free distance, the distance of the searched error vectors is updated to the free distance, then the old search is repeated, and the optimal adaptive spatial modulation having the updated free distance as a result of the repeated old search is performed. And transmitting the signal using the optimal adaptive spatial modulation set to the old search adaptive spatial modulation (SS-ASM) method. The method of claim 1,
The old search is performed in a sphere decoding receiver for generalized spatial modulation, and the old search adaptive spatial modulation method is characterized in that the search space is limited to the distance between error vectors having a single non-zero value with low complexity. . The method of claim 1,
The old search is the old search adaptive spatial modulation method characterized in that by setting the free distance to induce a minimum BER, and then the search early termination method is applied when the distance of the searched error vectors is smaller than the free distance. A multiple input / output (MIMO) system using the SS-ASM method according to any one of claims 1 to 3.

Images