KR101135230B1 - System for nulling interference using weighted precoding in mimo cognitive radio system - Google Patents

Abstract

Disclosed is an interference cancellation system using a weighted precoder in a cognitive radio system using multiple antennas. A weight allocator configured to calculate a weight to be applied to the precoder using the similarity between spatial fading correlation matrices in a cognitive radio system using multiple antennas; A power allocator for allocating power of the precoder using a weighted singular matrix; And a weight determining precoder including a beam determiner configured to determine a beam direction of the precoder using a singular matrix. The cognitive radio system is configured to transmit information through a ZF-based beamforming algorithm and a weighted precoder. Send.

Description

Interference Cancellation System Using Weighted Precoder in Cognitive Radio System Using Multiple Antennas {SYSTEM FOR NULLING INTERFERENCE USING WEIGHTED PRECODING IN MIMO COGNITIVE RADIO SYSTEM}

Embodiments of the present invention relate to a linear precoding technique for transmission between cognitive radio users, in which a cognitive radio system having multiple antennas does not interfere with the main user in an environment in which the cognitive radio system coexists with the main user. .

Recently, in order to increase the efficiency of a cognitive radio system, a method for increasing the transmission rate and the bit error rate (BER) performance of a cognitive radio system using a multi-antenna technique has been proposed. Multi-antenna technology can increase channel capacity through spatial multiplexing using limited frequency resources and transmit power in a channel environment without spatial correlation, and is a space-time block that is a coding technique that obtains transmit diversity gain. By applying Space Time Block Coding (STBC), the BER performance of the system can be improved.

However, in a real propagation environment, multiple antenna channels are correlated with each other due to low channel scattering, and channel capacity is lowered than when assuming that there is no spatial correlation. In order to solve this drawback, when channel information is used at the transmitter, a significant gain can be obtained in terms of channel capacity, bit error rate, and the like compared to techniques using channel information only at the receiver. In particular, even when only the spatial fading correlation matrix, which is partial channel information, is used, it is possible to implement a robust scheme for channel fading.

The transmission precoder of the cognitive radio system using the correlation matrix of the transmitting antennas, which is partial channel information, can be designed by applying the interference threshold to the main user and the maximum transmission power condition of the secondary user. have. Here, the correlation matrix of the transmitting antennas uses a matrix of the link between the sub-user transmitting end and the receiving end, and the link between the sub-user transmitting end and the main user receiving end, respectively.

As shown in FIG. 1, a cognitive radio system having a plurality of antennas in the transceiver 101 and a main user receiver 201 having one receive antenna are considered. In this case, it is assumed that N _T antennas of the transmitting terminal 101 of the cognitive radio system are correlated with each other, and that the antennas of the N _R sub-user receiving terminals 102 and the antennas of the main user receiving terminal 201 are not correlated with each other.

A cognitive radio system may be represented as shown in FIG. The randomly generated bits 301 are modulated 302 and then orthogonal space-time block coded 303. In this case, the space-time code C 304 passes through a linear precoder 306 having a complex value with a ZF technique and a weighted value using the antenna correlation information 305-1 and 305-2, and then transmits it through the MIMO channel 307. do. The information Y 308 received at the receiving end may be represented by Equation 1 below.

Where H represents a MIMO channel 307, F represents a linear precoding matrix 306, and C represents an orthogonal space time block coding (OSTBC) codeword 304, respectively. In addition, it is assumed that R _ts is a fading correlation matrix 305-1 in the sub-user MIMO link and N is a zero mean circularly symmetric complex Gaussian (ZMCSCG) having a variance of σ2.

The received information Y 308 is detected using the Maximum Likelihood (ML) detection scheme 309 and then demodulated 310.

Assuming that ML detection is used in a flat fading channel environment, the optimal precoder is to transmit in the non-zero eigenmode of the transmit antenna correlation matrix. The eigen mode transmission method of the transmission spatial fading correlation matrix is called unique beamforming. Inherent beamforming may be represented as in Equation 2.

은 워터 필링 방법을 통해 얻어진다. 여기에서 워터 레벨은 종래의 워터 필링 방법과는 달리 고유벡터에 따른 상수가 아니기 때문에 멀티 레벨(multi-level) 워터 필링 방법으로 불리며 할당된 전력은 수학식 3과 같이 나타낼 수 있다.Optimal beam direction U _Rts is the eigenvector of R _ts and optimal power allocation

Is obtained through a water peeling method. Here, since the water level is not a constant according to the eigenvector, unlike the conventional water filling method, the water level is called a multi-level water filling method and the allocated power may be expressed as Equation 3 below.

와

는 각각 인지무선 시스템의 제한 조건인 송신 전력 제한 조건, 주사용자로의 간섭 제한 조건과 관련된 음수가 아닌 라그랑지안 계수이다. u_i(R_ts)는 U_Rts의 i 번째 열벡터를 나타내고 λ_i(R_ts)는 λ_Rts의 i 번째 대각 요소를 나타낸다. In equation (3)

Wow

Are the non-negative Lagrangian coefficients associated with the transmit power constraints and the interference constraints to the main user, respectively. u _i (R _ts ) represents the i th column vector of U _Rts and λ _i (R _ts ) represents the i th diagonal element of λ _Rts .

By applying an interference nulling technique to the original beamforming as described above, the bit error rate performance and the interference suppression performance of the cognitive radio system can be significantly improved. Among these linear precoding techniques, there is a projected unique beamforming algorithm using ZF criteria. The projected eigenbeamforming algorithm assumes that U _Rts (k) represents k column vectors corresponding to eigenvalues up to the upper k th of U _Rts ( k ≤ N _T ), where R _ts is U _Rts (k) spanned. (span) projected into space N (U _Rts (k)). The projected correlation matrix of this eigenbeamforming algorithm can be expressed as Equation 4.

은 특이값 분해(Singular value decomposition; SVD)를 통해 수학식 5와 같이 나타낼 수 있다. This projected matrix

Can be represented by Equation 5 through singular value decomposition (SVD).

는 각각

의 좌측 특이 벡터, 우측 특이 벡터라고 하고

는

의 특이값을 포함하는 대각행렬이다. 특이값 분해를 통해 최적의 선형 프리코더는 수학식 6과 같이 나타낼 수 있으며 이때의 빔방향과 할당된 전력은 수학식 5에서 얻어진 우측 특이벡터와 대각행렬로 나타내어진다.Where V and

Respectively

Left singular vector of, right singular vector of

Is

Diagonal matrix containing singular values of. Through singular value decomposition, the optimal linear precoder can be represented by Equation 6, and the beam direction and the allocated power are represented by the right singular vector obtained by Equation 5 and the diagonal matrix.

는 단위 행렬 I 가 되므로 사영 상관 행렬

는 영행렬이 된다. 즉, k=N_T가 되었을 경우, 기존의 사영된 고유 빔포밍 프리코더를 얻을 수 없게 되는데 이것은 기존의 ZF 기반 빔포밍 기법이 최대 k=N_T-1에 해당하는 간섭만 제거시킬 수 있다는 것을 의미한다. 즉, 기존의 알고리즘은 최소 간섭 부공간에 대한 널링이 불가능하기 때문에 주사용자로의 간섭 전력을 줄이는데 한계가 있다.In the conventional ZF-based beamforming scheme, if k = N _T in Equation 4, U _Rts (k) becomes U _Rts . U _Rts is a unitary matrix

Becomes the identity matrix I, so the projective correlation matrix

Becomes a zero matrix. In other words, when k = N _T , the existing projected unique beamforming precoder cannot be obtained, which indicates that the existing ZF-based beamforming technique can only remove interference up to k = N _T -1. it means. That is, the existing algorithm has a limitation in reducing the interference power to the main user because it is impossible to null the minimum interference subspace.

As a precoding means for solving the disadvantage that the conventional ZF technique beamforming algorithm does not apply nulling for the minimum interference subspace, a linear precoding technique is applied, which is weighted based on the vector similarity with the ZF technique.

In an environment in which a cognitive radio system having multiple antennas coexists with a main user, an interference cancellation system through a weighted precoder is provided to reduce the interference power to the main user and improve the performance of the cognitive radio system.

An interference cancellation system using a precoder in a cognitive radio system using multiple antennas, comprising: a weight assignment unit for calculating a weight to be applied to the precoder using the similarity between spatial fading correlation matrices in the cognitive radio system; A power allocator for allocating power of the precoder using a weighted singular matrix; And a beam determiner configured to determine a beam direction of the precoder using the singular matrix.

According to one side, the weight allocator may calculate the similarity between the spatial fading correlation matrix by a method of projecting one vector into another vector or interference subspace.

According to another aspect, the weight allocator assigns weights to the singular vectors derived through the singular value decomposition of the spatial fading correlation matrix.

According to another aspect, a cognitive radio system weights a zero-forcing (ZF) based beamforming algorithm to pass information through a multiple input multiple output (MIMO) channel through a ZF based beamforming algorithm and a weighted precoder. Send.

According to another aspect, an interference cancellation system using a weighted precoder applies a ZF-based beamforming algorithm to remove a vector of a spatial fading correlation matrix corresponding to interference to a main user through projection into an interference subspace. Further comprising a nulling unit, the precoder may apply a weight to remove or reduce the interference component that is not removed from the nulling unit.

By applying weights derived from vector similarity to ZF-based beamforming algorithms, interference suppression performance can be improved by reducing the interference power to the main user and improving the performance of bit error rate and channel capacity of cognitive radio system. Can be.

1 is a diagram illustrating a cognitive radio MIMO channel model.
2 is a block diagram illustrating an internal configuration of a cognitive radio MIMO system.
3 is a block diagram illustrating a precoder configuration to which weights are applied in a cognitive radio system using multiple antennas according to the present embodiment.
4 is a diagram for explaining the projection principle to the subspace in this embodiment.
FIG. 5 is a diagram illustrating an algorithm for calculating a weight based on the similarity of vectors in the present embodiment.
6 is a flowchart illustrating a precoding process using weights in a cognitive radio system using multiple antennas according to the present embodiment.
7 to 12 are diagrams for explaining the effects that can be obtained by weighted precoding in the present embodiment.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

3 is a block diagram illustrating an internal configuration of an interference cancellation system using a precoder to which weights are applied in a cognitive radio system using multiple antennas according to the present embodiment. FIG. 3 illustrates a linear precoder 506 in a cognitive radio system with complex values with ZF techniques and weights using antenna correlation information 505-1, 505-2.

In the interference cancellation system using the weighted precoder according to an embodiment, the randomly generated bits 501 are modulated by the modulator 502 and are orthogonal space-time block coded through the STBC encoder 503. In this case, the space-time code (C) 504 encoded by the STBC encoder 503 is a linear precoder 506 having a complex value to which the ZF technique and the weight are applied using the antenna correlation information 505-1 and 505-2. After passing through the MIMO channel 507 is transmitted. The information (Y) 508 received at the receiving end is expressed by Equation 1 below. The received information 508 is detected by the ML decoder 509 using the ML detection scheme and then demodulated by the demodulator 510.

As shown in FIG. 3, the linear precoder 506 of the cognitive radio system according to an embodiment includes a nulling unit 506-1, a weight allocation unit 506-2, and a power allocation unit 506-3. , The beam determination unit 506-4.

In the present exemplary embodiment, the linear precoder 506 having a complex value with a ZF technique and a weighted value may include a nulling unit 506-1 and a null that remove an interference component corresponding to a transmission direction to a main user using the ZF technique. A weight allocator 506-2 for allocating weights to reduce interference not removed by the ring unit 506-1, a power allocator 506-3 for allocating power, and a beam for determining information in a beam direction to transmit information And a decision unit 506-4.

(403)으로 사영되는 것을 나타낸다. 여기에서,

(401-2)는 A(401-1)의 사영된 벡터를 나타내고,

(403)는 벡터

(402-1)와

(402-2)에 의해 스팬된(span) 부공간을 나타낸다. 그리고,

(404)는 벡터 A(401-1)와 부공간

(403)사이의 각을 나타낸다. 도 4의 (A)는

(404)가 큰 경우, 도 4의 (B)는

(404)가 작은 경우이다. 도 4에서 사영된 벡터는 수학식 7과 같이 나타낼 수 있다. 4 is a diagram for explaining the projection principle to the subspace in this embodiment. As can be seen from the dot product of the vector, the similarity of the vector direction can be determined by projecting one vector into another vector or subspace. 4 A (401-1), U A ( 401-1) , when said (402-1) is the same of different vector direction size of the subspace

It is projected to (403). From here,

401-2 represents the projected vector of A (401-1),

403 is a vector

(402-1) and

The subspace spanned by (402-2). And,

404 is a vector A (401-1) and a subspace

An angle between 403 is shown. 4A is

If 404 is large, Fig. 4B is

404 is small. The projected vector in FIG. 4 may be represented as in Equation 7.

는 점점 작아질 것이고 벡터 A와 사영된 벡터

의 크기(norm)비는 1에 가까워질 것이다. 이를 수학식 8과 같이 나타낼 수 있다.As shown in Fig. 4, if the difference in the directions of the vectors A 401-1 and U 402-1 becomes small, the angle between the two vectors is small.

Will be getting smaller and the projected vector A and

The norm ratio will be close to one. This may be represented as in Equation 8.

That is, the similarity between the vector and the vector forming the projective subspace can be determined by the norm ratio between the vector and the projected vector. By applying this basic principle, weights based on the similarity between the interference subspace and the cognitive radio system spatial fading correlation matrix can be derived. In addition, weights are applied to the singular vectors derived through singular value decomposition of the spatial fading correlation matrix in order to reduce interference that is not eliminated in the nulling process.

)이 주사용자로의 간섭 방향에 해당하는 최적의 빔 방향(

)의 열벡터와 방향이 유사해진다면

와 사영된

의 크기비는 1에 가까워질 것이다. 이것을 이용하여 주사용자로의 간섭을 제거하기 위한 가중치를 계산할 수 있다.In other words, the fading correlation matrix in the secondary user MIMO link (

) Is the optimal beam direction (

If the direction is similar to the column vector of

Projected with

The size ratio of will be close to one. This can be used to calculate weights to eliminate interference to the main user.

이다. 위 가중치를 적용한 선형 프리코딩의 전력 할당에 관한 수식은 수학식 9, 수학식 10과 같이 나타낼 수 있다.FIG. 5 is a diagram illustrating an algorithm for calculating a weight based on the similarity of vectors in the present embodiment. In FIG. 5, the matrix W corresponding to the weight is used to apply the weight without increasing the transmission power.

to be. Equation regarding the power allocation of the linear precoding to which the above weight is applied may be expressed as Equation 9 and Equation 10.

는

의 우측 특이 벡터이고 W는 N_T개의 음수가 아닌 값을 가지는 N_T

N_T의 대각행렬이다. 라그랑지안 계수는 이진 검색(Binary search) 알고리즘을 사용하여 구한다. 최종적으로 제안된 선형 프리코딩 알고리즘은 도 5를 통해 설명한 알고리즘과 수학식 11과 같이 나타낼 수 있다. 수학식 12과 수학식 13은 인지무선 간섭 제한 조건들로

는 부사용자의 최대 전송 전력,

는 주사용자로의 간섭 임계값을 나타낸다. here,

Is

Right singular vectors of W is a N _T N _T has a value which is not a negative one

Diagonal matrix of N _T. Lagrangian coefficients are obtained using a binary search algorithm. Finally, the proposed linear precoding algorithm may be represented as shown in Equation 11 and the algorithm described with reference to FIG. 5. Equations 12 and 13 are cognitive radio interference constraints.

Is the maximum transmit power of the secondary user,

Represents an interference threshold to the main user.

6 is a flowchart illustrating a precoding process using weights in a cognitive radio system using multiple antennas according to the present embodiment. An interference cancellation method using a weighted precoder according to an embodiment is executed by the interference cancellation system described with reference to FIG. 3.

In step 501, the interference elimination system using the weighted precoder removes the interference component corresponding to the transmission direction from the space-time code C encoded through the ZF scheme to the main user using the antenna correlation information 500.

In steps 502 and 503, the interference cancellation system using the weighted precoder calculates the singular vectors of the projected spatial fading correlation matrix 507 and the spatial fading correlation matrix 507 through singular value decomposition (SVD). .

In step 504, the interference cancellation system using the weighted precoder may calculate the weight using the similarity between the spatial fading correlation matrices to reduce interference that was not removed in step 501.

In step 505, the interference cancellation system via the weighted precoder calculates the power to be allocated to the precoder based on the weighted singularity matrix 508.

In step 506 the interference cancellation system via the weighted precoder determines the beam direction of the precoder based on the singular matrix 508.

According to the above process, a cognitive radio system having multiple antennas applies a weight to a ZF-based beamforming algorithm and passes a weighted precoder based on the similarity between the ZF-based beamforming algorithm and a vector to a multiple input multiple output. Information can be transmitted through the channel.

7 is a graph showing the average interference power level at the main user when the interference threshold is -10dB and the angle spread (AS) is 25 °. As shown in FIG. 7, it can be seen that the proposed algorithm exhibits excellent interference suppression performance of 2 to 8 dB over the entire SNR (signal-to-noise ratio) region.

FIG. 8 is a projected inherent beamforming nulling only one interference subspace under the condition of an angle spread of 25 ° and an interference threshold of -10 dB, and the SER (symbol error probability) of the proposed algorithm. Graph showing performance.

FIG. 9 is a graph showing eigenbeamforming under angular spread of 25 ° and interference threshold of -15dB, projected eigenbeamforming nulling only one interfering subspace, and SER performance of the proposed algorithm. .

10 and 11 are graphs showing the eigen beamforming, the projected eigenbeam beamforming nulling a plurality of interference subspaces, and the SER performance of the algorithm proposed in the present invention when the interference thresholds are -10 dB and -15 dB, respectively. to be.

As shown in FIGS. 7 to 11, the higher the SNR or the lower the interference threshold, the better the performance of the algorithm proposed in the present invention.

12 is a graph comparing channel capacity of each algorithm according to SNR when the interference threshold is -10dB and the AS is 25 °. Overall, the algorithm proposed in the present invention, which has better interference suppression capability than the existing algorithm, has better performance in terms of transmission capacity as the interference condition becomes more severe.

In the cognitive radio system according to an embodiment, the interference elimination system using the precoder applying the weight is applied to the weighting method according to the similarity of the vector. Even small interference components can be removed by weight, thus reducing the amount of interference that has not been eliminated through conventional algorithms. In addition, since the weight is applied to the interference subspace that is not removed, the interference suppression performance can be improved by suppressing the interference by the applied weight.

As described above, the present invention has been described by way of limited embodiments and drawings, but the present invention is not limited to the above embodiments, and those skilled in the art to which the present invention pertains various modifications and variations from such descriptions. This is possible.

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.

502: modulator
503: STBC encoder
506: linear precoder
506-1: knurled part
506-2: weight allocation unit
506-3: power allocation unit
506-4: beam determination unit
507: MIMO channel
509: ML decoder
510: demodulator

Claims (6)

In the interference cancellation system using a precoder in a cognitive radio system using multiple antennas,
A weight allocator configured to calculate a weight to be applied to the precoder using the similarity between spatial fading correlation matrices in the cognitive radio system;
A power allocator for allocating power of the precoder using the weighted singular matrix; And
A beam determination unit which determines a beam direction of the precoder using the singular matrix
Including,
The weight allocation unit,
Calculating the similarity between the spatial fading correlation matrices by projecting one vector into another vector or interference subspace
Interference cancellation system using a weighted precoder characterized in that. delete The method of claim 1,
The weight allocation unit,
Assigning the weights to a singular vector derived through singular value decomposition of the spatial fading correlation matrix
Interference cancellation system using a weighted precoder characterized in that. The method of claim 1,
The cognitive radio system,
Applying the weight to a zero-forcing (ZF) based beamforming algorithm to transmit information through a multiple input multiple output (MIMO) channel through the ZF based beamforming algorithm and the weighted precoder
Interference cancellation system using a weighted precoder characterized in that. Following claim 1
The interference cancellation system using the weight precoder,
A nulling unit that removes the vector of the spatial fading correlation matrix corresponding to the interference to the main user through projection to the interference subspace by applying a ZF-based beamforming algorithm.
Interference cancellation system through a weighted precoder further comprising. The method of claim 5,
The precoder,
Applying the weight to remove or reduce interference components that are not eliminated in the nulling portion
Interference cancellation system using a weighted precoder characterized in that.

Images