KR101094442B1 - Adaptive interference alignment method for time-varying multiuser mimo interference channels - Google Patents

Abstract

The present invention relates to a method for obtaining a beamforming vector for efficiently performing interference alignment. The technical problem to be solved is to solve the eigenvalues obtained in the process of obtaining a previous beamforming vector without performing a complicated algorithm for obtaining the beamforming vector at each instant. The present invention provides a simple method for obtaining a new beamforming vector using only eigenvectors and channel differences.

To this end, the interference alignment method according to the present invention calculates a beamforming vector using an iterative least square method after predicting the first downlink channel, and eigenvalues and eigenvalues obtained through the first downlink channel until the channel change is not severe. The new beamforming vector is calculated by using the vector and Equation 15, and the time-varying multi-user multi-antenna is characterized by obtaining a new beamforming vector based on the predicted channel when the channel change becomes severe after repeating the above calculation process. An adaptive interference alignment method in an interference channel environment is disclosed.

Beamforming vector, Kalman filter, interference alignment method, time division duplex system

Description

ADAPTIVE INTERFERENCE ALIGNMENT METHOD FOR TIME-VARYING MULTIUSER MIMO INTERFERENCE CHANNELS

The present invention relates to a method for obtaining a beamforming vector for efficiently performing interference alignment in an interference alignment technique for dividing a desired signal and interference into different signal spaces.

In cellular communication, a service area is divided into 'cells', which are small regional units. Such 'cells' mean an area covered by a base station. The size of the 'cell' may be changed according to the density of the subscriber. In general, as the density of the subscriber increases, the size of the cell decreases.

Therefore, the area in which the 'cells' overlap is continuously increasing, and in this overlapping area, system performance is degraded due to interference between signals. An interference alignment technique has been proposed as an effective solution for such interference effects.

The interference alignment technique is a transmission beamforming technique in which all interference signals exist in a limited interference space by dividing the entire space into a desired signal space and an interference signal space on the assumption that all direct links and channels of the interference link are known.

If the desired signal space and the interference signal space are linearly independent, the desired signal can be extracted without the influence of interference, so that a desired degree of freedom, that is, a multiplexing gain can be obtained.

On the other hand, when the channel does not change over time, the beamforming vector obtained initially can be used continuously. However, when the channel changes over time, the performance deterioration is severe when such a method is used.

Therefore, it is necessary to predict a channel to change with each transmission time and to calculate a new beamforming vector at every moment according to the predicted channel.

However, since it is complicated to find a new beamforming vector every moment, a method for calculating a new adaptive beamforming vector is required.

However, all of the interference alignment techniques proposed to date are related to a method of calculating a beamforming vector for interference alignment under the assumption that channel information is known correctly. The research on the situation is not progressed yet.

Of course, a method of newly calculating an interference alignment beamforming vector based on channel values changed at every instant of channel change can be considered, but this method has a problem of high complexity and unreality.

The present invention was invented to solve the above problems, and estimates an uplink channel based on an uplink pilot in a time division duplex (TDD) system, and based on the estimated channel, uplink An object of the present invention is to provide an interference alignment method of improving performance by predicting a channel of a link by reducing a difference between a channel used for interference alignment and a channel through which an actual transmission signal passes.

In addition, the interference alignment simply finds a new beamforming vector using only the eigenvalues, the eigenvectors, and the channel difference obtained in the process of obtaining the previous beamforming vector without performing a complicated algorithm for finding the beamforming vector at each moment using the predicted channel. The purpose is to provide a method.

In order to achieve the above object, the interference alignment method according to the present invention calculates a beamforming vector using an iterative least square method after predicting the first downlink channel, and the first downlink until the channel change is not severe. The new beamforming vector is calculated using the eigenvalues, the eigenvectors, and Equation 15 obtained through the channel, and if the channel change becomes severe after repeating the above calculation process, a new beamforming vector can be obtained based on the predicted channel again. have.

In addition, when the new beamforming vector period is L based on the moving speed of the terminal, the beamforming matrix can be obtained according to algorithm 1.

인 단계에서도 A_ij에 대한 갱신을 수행할 수 있다.Also, unlike Algorithm 1, the update on A _ij is performed only in every L stage.

In the step A _ij can be updated.

와 H_ij[n]을 가지고 하기의 수학식 7을 계산하여 v[n]을 구하는 제 1과정과, 상기 제 1과정에서 구한 v[n]과 H_ij[n]을 이용하여 하기의 수학식 10을 이용하여

을 다시 갱신하는 과정을 수렴할 때까지 반복 수행하여 최종

를 구하는 제 2과정과, 상기 제 2과정에서 구한

와 현재 시간의 예측 채널 행렬 H_ij[n]을 이용하여

을 구성하는 제 3과정과,

을 고유치 분해하여 고유치

와 고유치 벡터

을 구하는 제 4과정과, v[n]=q₁으로 결정하여 현재 시간 n=kL에서의 빔형성 벡터로 사용하고, 나머지 고유벡터와 고유치는 다음 주기 전까지(n=kL+1,...,kL+L-1) 빔형성 벡터 계산을 위해서 저장해 두는 제 5과정과, n=kL 단계에서 구해 놓은

벡터들과

및

을 이용하여 n=kL+1,...,kL+L-1 시간에서 반복 계산없이 빔형성 벡터를 구하는 제 6과정을 포함한다. In addition, Algorithm 1 is initialized at the time n = kL, k = 0, 1...

And H _ij [n] to calculate v [n] by calculating Equation 7 below, and using v [n] and H _ij [n] obtained in the first step Using 10

To repeat the process of renewing

The second process of obtaining the, and obtained in the second process

And the prediction channel matrix H _ij [n] of the current time

The third process constituting the

Eigenvalues

And eigenvalue vectors

In the fourth process of finding, we determine v [n] = q ₁ and use it as a beamforming vector at the current time n = kL, and the remaining eigenvectors and eigenvalues before the next period (n = kL + 1, ... , kL + L-1) The fifth step is stored for the beamforming vector calculation, and n = kL

With vectors

And

The sixth step of obtaining a beamforming vector without iterative calculation at time n = kL + 1, ..., kL + L-1 using

In addition, the adaptive interference alignment method described above may be employed in a data transmitter in a time varying multi-user multi-antenna interference channel environment.

As described above, according to the adaptive interference alignment method in the time-varying multi-user multi-antenna interference channel environment according to the present invention, in the time division multiplexing system, the uplink channel is estimated based on the uplink pilot, and the estimated channel is based on the estimated channel. By predicting the channel of the uplink, it is possible to reduce the difference between the channel used for interference alignment and the channel through which the actual transmission signal passes, thereby improving performance.

In addition, a new beamforming vector can be easily obtained using only the eigenvalues, the eigenvectors, and the channel differences obtained in the process of obtaining the previous beamforming vector without performing a complicated algorithm for finding the beamforming vector at each moment using the predicted channel. It works.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. In describing the present invention, detailed descriptions of related well-known functions or configurations are omitted in order not to obscure the gist of the present invention.

The present invention is based on an interference alignment technique. Therefore, consider the system configuration as shown in FIG. 1 illustrates a K-user MIMO interference channel environment in which a transmitter and a receiver having K antennas exist. That is, it is assumed that there are K transmitters and receivers corresponding to each transmitter, and each transmitter and receiver has N and M antennas. In addition, the signal received at the nth time from the receiving end k may be expressed by Equation 1 below due to the characteristics of the wireless channel.

Here, s _k [n] and V _k [n] represent a transmission data vector (d × 1) and a beamforming matrix (N × d) transmitted by the transmitter k. H _kl [n] and n _k [n] indicate a channel matrix M × N and a noise vector M × 1 from the transmitter l to the receiver k. Thus, the first term of Equation 1 represents a 'desired signal' and the second and third terms represent 'interference' and 'noise'.

In the meantime, the interference alignment technique divides the M-dimensional signal space into the d-dimensional signal subspace and the (Md) -dimension subspace to align (K-1) interference signals in the (Md) -dimensional space. By means of securing a signal space independent of the interference space means a technique that completely eliminates the influence of the interference in the desired signal.

이다. 이와 같은 최대 자유도를 얻는 경우는 각 송신단에서

개의 데이터 스트림을 전송하는, 즉

인 경우이다. 이와 같은 경우 간섭 정렬 조건을 하기의 수학식 2와 같이 나타낼 수 있다.The maximum degree of freedom in this way is

to be. When the maximum degree of freedom is obtained, each transmitting end

To send data streams, i.e.

If In such a case, the interference alignment condition may be expressed by Equation 2 below.

Here, C (A) means a column space in which columns of the matrix A span. Accordingly, each row of Equation 2 represents a space of signals to be aligned at each receiving end. Under the above conditions, the first equation of the first row may be represented as in Equation 3 below.

는 V_i의 k번째 열을 의미하고,

은 선형 조합 계수이다. 즉, 두 행렬이 스팬(span)하는 열공간이 일치한다는 것은 한 쪽 행렬의 각 열은 다른 쪽 행렬의 선형 조합으로 나타낼 수 있다는 것을 의미한다, 이를 크로네커 곱(Kronecker Product)을 이용하여 다시 정리하면 하기의 수학식 4와 같이 나타낼 수 있다.here,

Is the kth column of V _i ,

Is a linear combination coefficient. In other words, the coincidence of the column space spanning the two matrices means that each column of one matrix can be represented as a linear combination of the other matrix, which can be rearranged using the Kronecker product. It can be expressed as Equation 4 below.

는 두 행렬의 크로네커 곱(Kronecker Product)이다. 그리고, vec(A)은 행렬의 열을 계속 쌓아서 만든 벡터를 의미한다.Where I _d represents a d × d unit matrix,

Is the Kronecker product of two matrices. And, vec (A) means a vector made by stacking the columns of the matrix.

Given A _ij and H _ij based on the above results, the entire equation of Equation 2 can be represented by one linear equation as shown in Equation 5 below.

를 k번째 행과 l번째 열로 갖는 선형 조합 계수 행렬이다. 이 때 앞에 있는 행렬을

라고 정의하고 뒤에 있는 벡터를 v로 정의하면 하기의 수학식 6과 같이 나타낼 수 있다.Where A _ij is

Is a linear combination coefficient matrix with k th row and l th column. In this case, the matrix

If it is defined as and the vector after the v can be expressed as in Equation 6 below.

행렬의 크기는

이다. 따라서 사용자의 수 및 안테나의 수에 따라서 overdetemined가 될 수 있다. 예를 들어, K≥4이고, M=N인 경우는 overdetermined가 되는 경우이다. 이와 같은 경우에는 상기의 수학식 6을 만족하는 해가 존재하지 않기 때문에 근사해를 찾아야 한다. 그래서 일반적으로 사용하는 최소 자승법(Least Squares, LS)을 이용하여 근사해를 찾는다.here

The size of the matrix

to be. Therefore, it can be overdetemined depending on the number of users and the number of antennas. For example, K≥4 and M = N is a case where it becomes overdetermined. In this case, since there is no solution that satisfies Equation 6 above, an approximate solution must be found. So we find the approximate solution using the least-squares method (LS) that is commonly used.

의 가장 작은 고유치에 해당하는 고유벡터이다. 그리고, 이렇게 최적의

을 구한 후에는 이 값을 이용하여 다시 최적은 A_ij들을 계산할 수 있다. 우선 A₁₃을 구하는 과정을 살펴보면 다음과 같다.The solution of Equation 7 is

Eigenvectors corresponding to the smallest eigenvalues of. And so optimal

After obtaining, we can use this value to calculate optimal A _ij again. First, the process of obtaining A ₁₃ is as follows.

First, Equation 3 may be rearranged as in Equation 8 below.

Here, when H _ij [n] and v [n] are given, A ₁₃ can be obtained as shown in Equation 9 below.

The equation for obtaining A _ij in this manner is summarized as in Equation 10 below.

Accordingly, when it is repeated until the process of obtaining v [n] using Equation 7 and the process of obtaining A _ij using Equation 10 is converged, the final interference alignment beamforming vector v [n]. Can be obtained.

On the other hand, if the channels H _ij [n] are always the same regardless of the time n, the beamforming vector obtained by performing the above-described process once can be used. However, when the channel is changed, there is a difference between the channel used for calculating the beamforming vector and the actual channel, thereby preventing interference alignment from being properly performed. Therefore, in the time-varying channel, a channel prediction process is required to minimize such channel difference.

If the entire frame structure is given as shown in FIG. 2 in a time division duplex (TDD) system, the uplink channel and the downlink channel may be the same. Therefore, the channel estimated by the uplink can be regarded as the channel of the downlink, and since the pilot symbol exists in the uplink period, the channel can be estimated. In addition, an optimal channel estimate may be obtained using a Kalman filter under the assumption that the channel operates like the AR model (AutoRegressive model).

However, in the downlink period, there is no reception pilot since it is a transmission period. Therefore, channel prediction must be performed based on a previous uplink channel, and this process is represented in the form of a graph model as shown in FIG. 3.

Since the channel estimation and prediction process is the same regardless of the transmit / receive end factor or antenna factor, all the factors are omitted. Therefore, when the channel is modeled as the m-th order AR process (AutoRegressive process) can be expressed as shown in Equation 11 below.

Here, the variance of the coefficient β [p] and Gaussian Noise u [n] can be obtained through the Yule-Walker equation. Using this, the channel and the reception pilot may be represented in Equation 12 below in the uplink period.

라 하고, 이 때 프로세스 잡음 벡터를

라 정의한다. 여기서 z[n]은 수신 파일럿 신호를 나타낸다. 따라서

이고, w[n] 파일럿 수신 과정의 잡음 즉 측정 잡음(measurement noise)이다. Where a channel vector consisting of channels from nth to n-m + 1st time

Where the process noise vector

It is defined as Where z [n] represents the received pilot signal. therefore

W [n] is the noise of the pilot reception process, that is, measurement noise.

In this way, when a state-space model is given, the Kalman filter is given by Equation 13 below.

과

과 같이 주어지고, k와

는 칼만 이득(Kalman Gain)과 평균 자승 오류(Mean Square Error) 행렬이다. Where the initial condition is

and

Is given by k and

Is the Kalman Gain and Mean Square Error matrix.

과

는 각각 프로세스 잡음과 측정 잡음의 분산을 나타낸다. 이와 같은 과정을

까지 수행한다. 그리고 상향 링크 구간의 채널 예측은 하기의 수학식 14와 같이 표현할 수 있다.

and

Denotes the variance of process noise and measurement noise, respectively. This process

Do until. And the channel prediction of the uplink interval can be expressed as shown in Equation 14 below.

이다.here,

to be.

On the other hand, v [n] may be newly calculated at every moment by using the channel H _ij [n] predicted in the above-described manner. However, since this calculation process is an iterative operation, the complexity is high, and during the desired time, Sometimes solutions may not be available. Therefore, the present invention proposes an adaptive interference alignment method that simplifies this process.

를 더하는 형태가 될 수 있다. 즉, 이전 빔형성 벡터에서 채널 변화량에 따른 빔형성 벡터 변화량을 더한 구조가 될 수 있다. H_ij[n-1]와 A_ij[n-1]가 주어진 상황에서 v[n]는

의 최소 고유치에 해당하는 고유벡터이다. 그리고, H_ij[n]와 A_ij[n]가 주어진 상황에서 v[n]는

의 최소 고유치에 해당하는 고유벡터이다. 이를 바탕으로 하기의 수학식 15와 같은 관계를 찾을 수 있다.When a channel change is not fast beamforming obtained using the H _ij [n] vector v [n] is calculated by using the beamforming H _ij [n-1] vector v [n-1]

It can be in the form of adding. That is, the structure may be obtained by adding the beamforming vector change amount according to the channel change amount in the previous beamforming vector. Given H _ij [n-1] and A _ij [n-1], v [n] is

Eigenvectors corresponding to the minimum eigenvalues of. And, given H _ij [n] and A _ij [n], v [n] is

Eigenvectors corresponding to the minimum eigenvalues of. Based on this, the relationship as shown in Equation 15 below can be found.

이고,here,

ego,

과,

는

의 오름차순으로 정렬된 고유치와 그에 해당하는 고유벡터들이다.

and,

Is

Eigenvalues arranged in ascending order of and corresponding eigenvectors.

의 고유치와 고유벡터 그리고 채널 차이

만 주어지면 새로운 채널에 해당하는 빔형성 벡터 v[n]을 구할 수 있다. 이와 같은 관계를 이용하여 시변 채널 환경에 적합한 적응형 간섭 정렬 알고리즘을 만들 수 있다.In other words, a matrix of previous time channels

Eigenvalues, Eigenvectors, and Channel Differences of

If only given, the beamforming vector v [n] corresponding to the new channel can be obtained. This relationship can be used to create an adaptive interference alignment algorithm suitable for time-varying channel environments.

First, after predicting the first downlink channel, the beamforming vector is calculated by using the existing iterative least square method, and until the channel change is not severe, the equation 15 and In the same way, a new beamforming vector is obtained. Next, if the channel change becomes severe while repeating the above process, the beamforming vector is newly obtained based on the predicted channel. When the new beamforming vector period is L based on the moving speed of the terminal, the method as described above is summarized as in Algorithm 1 below.

Algorithm 1

-

를 이용하여

를 구성하고 v[n]과

를 구하는 과정을 수렴할 때까지 반복 수행하면서 최종

를 구한다.The predicted H _ij [n] and

Using

Configure v [n] and

Iterate over the convergence process to find the final

.

를 이용하여

을 구성한다.The predicted H _ij [n] and determined above

Using

Configure

을 고유치 분해하여

와

을 구한다.-

Eigenvalue decomposition

Wow

.

Reconstruct v [n] with v [n] = q ₁ to obtain the beamforming matrix v _i [n] for each transmitter.

for n = kL + 1, ..., (i + 1) L-1

를 이용하여

을 구성하고, G_iL[n]을 계산한다.The predicted H _ij [n] and determined above

Using

And calculate G _iL [n].

-V [n] is obtained by using Equation 16 below.

Reconstruct v [n] with v [n] = q ₁ to obtain the beamforming matrix V _i [n] for each transmitter.

인 단계에서 수행할 수도 있다.In Algorithm 1, the update on A _ij is performed only in every L stage, but if the complexity is allowed,

It may also be performed in the in stage.

Hereinafter, the algorithm 1 will be described in more detail.

First, time goes by n = 0, 1, 2, 3, ..., L, L + 1, ..., 2L, 2L + 1, ..., 3L, ..., It is _assumed that the MIMO channel matrix H _ij [n] corresponding to time is predicted. That is, given H _ij [n], n = 0, 1, 2, 3, ..., L, L + 1, ... 2L, 2L + 1, ..., 3L, ..., It is an object of the present invention to obtain a beamforming vector corresponding to each time n in.

와 H_ij[n]을 가지고 상기 수학식 7을 풀어서 v[n]을 구하고(제 1과정) 이렇게 구한 v[n]과 H_ij[n]을 가지고 수학식 10을 이용하여

을 다시 갱신하는 과정을 수렴할 때까지 반복 수행하여 최종

을 구한다(제 2과정). 이렇게 구한

와 현재 시간의 예측 채널 행렬 H_ij[n]을 이용하여 아래와 같이 정의되는

을 구성한다(제 3과정).To this end, L, which represents the algorithm period, is determined according to the channel change rate. In other words, if the channel change is fast, L is reduced to prevent performance degradation due to channel change, and in the opposite case, L is increased to reduce the complexity of the algorithm. So at time n = kL, k = 0,1 ..., the beginning of the algorithm's operating cycle,

And H _ij [n] to solve the above equation (7) to obtain v [n] (step 1) and using the obtained v [n] and H _ij [n] using equation (10)

To repeat the process of renewing

(Step 2). So obtained

And using the prediction channel matrix H _ij [n] of the current time

(Step 3).

을 고유치 분해 하여 고유치

와 고유치 벡터

을 구한다(제 4과정). 그래서 v[n]=q₁으로 결정하여 현재 시간 n=kL에서의 빔형성 벡터로 사용하고, 나머지 고유벡터와 고유치는 다음 주기 전까지, 즉 n=kL+1,...,kL+L-1, 빔형성 벡터 계산을 위해 저장해 둔다(제 5과정).And

Eigenvalues

And eigenvalue vectors

(Step 4). Therefore, we determine v [n] = q ₁ and use it as the beamforming vector at the current time n = kL, and the remaining eigenvectors and eigenvalues are before the next period, that is, n = kL + 1, ..., kL + L- 1, it is stored for the calculation of the beamforming vector (step 5).

벡터들과

및

을 이용하여 n=kL+1,...,kL+L-1 시간에서 빔형성 벡터를 반복 계산(iteration)없이 간단하게 구할 수 있다(제 6과정).This is obtained from the step n = kL

With vectors

And

The beamforming vector can be obtained simply without iteration at time n = kL + 1, ..., kL + L-1 using step 6 (step 6).

In order to show the performance of the proposed method, computer simulations are performed in the following environments. First, it is assumed that there are three transmitters and a receiver, and that the transmitter has four antennas and the receiver has two antennas. In general, such an antenna configuration is appropriate because the transmitting end is the base station and the receiving end is the terminal. The carrier frequency is set to 2.3 GHz and the symbol period is set to 115.2 μs. One frame consists of an uplink period and a downlink period each consisting of 25 symbols. And the time-varying channel was modeled by the 2nd AR model.

In this environment, when the moving speed of the terminal is changed from 1 km / h to 10 km / h, the beamforming vector is calculated using the channel estimated from the last symbol of the uplink without considering the time-varying channel and applied to all the downlinks. As the moving speed increases, the channel difference becomes larger and worsens the performance. This can be confirmed through the graph shown in FIG. 4. Therefore, even if the moving speed is only 10km / h, the throughput sum can be reduced by 50%.

However, the performance of predicting a downlink channel and determining a beamforming vector using the predicted channel, as in the method of the present invention, can be seen in the graph shown in FIG. 5. In the figure, 'direct' is a case of continuously calculating the beamforming vector using the predicted channel, and 'adaptive' is L = n _d in the proposed algorithm. Performance when set to = 25. First, we can see that the performance difference between the 'direct' and 'adaptive' methods is negligibly small, and the performance of the channel prediction is almost undegraded up to 5 km / h of travel speed. It can be seen that there is a decrease in performance. This degradation can be overcome by reducing n _d .

As described above with reference to the drawings illustrating an adaptive interference alignment method in a time-varying multi-user multi-antenna interference channel environment according to the present invention, the present invention is not limited to the embodiments and drawings disclosed herein. Of course, various modifications may be made by those skilled in the art within the technical scope of the present invention.

1 is a conceptual diagram illustrating a K-user MIMO interference channel environment in which a transmitter and a receiver having K antennas exist.

2 is a diagram illustrating a frame structure of a TDD system assumed in the present invention.

3 is a diagram illustrating a channel estimation process in an uplink period and a channel prediction process in a downlink based on a Kalman filter.

4 is a graph showing performance degradation according to a moving speed when three users are used, the number of transmitting antennas is four, and the number of receiving antennas is two.

FIG. 5 is a graph showing the performance of data rate sum when the beamforming vector for interference alignment is changed at every moment using the channel prediction result as in the method proposed in the present invention.

Claims (10)

A method for obtaining a beamforming vector for performing efficient interference alignment in a time varying multi-user multi-antenna interference channel environment, Calculating a beamforming vector using an iterative least square method after predicting the first downlink channel; Calculating a new beamforming vector using the eigenvalue and eigenvector obtained through the first downlink channel and Equation 15 below; [Equation 15]

(here,

ego,

and,

Quot;

Eigenvalues and their corresponding eigenvectors, sorted in ascending order of.) Computing the new beamforming vector, When the new beamforming vector period based on the moving speed of the terminal is called L, the beamforming matrix is obtained according to Algorithm 1 below. Algorithm 1 -

The predicted H _ij [n] and

Using

Configure v [n] and

Iterate over the convergence process to find the final

. The predicted H _ij [n] and determined above

Using

Configure -

Eigenvalue decomposition

Wow

. Reconstruct v [n] with v [n] = q ₁ to obtain the beamforming matrix v _i [n] for each transmitter. for n = kL + 1, ..., (i + 1) L-1 The predicted H _ij [n] and determined above

Using

And calculate G _iL [n]. -V [n] is obtained by using Equation 16 below. [Equation 16]

Reconstruct v [n] with v [n] = q ₁ to obtain the beamforming matrix V _i [n] for each transmitter. delete The method according to claim 1, Unlike in Algorithm 1, the update on A _ij is performed only in every L stage.

Adaptive interference alignment method in a time-varying multi-user multi-antenna interference channel environment, characterized in that for updating the A _ij . The method according to claim 1, Initially at time n = kL, k = 0, 1 ..., which is the beginning of the operation period of Algorithm 1

And H _ij [n] to calculate v [n] by calculating Equation 7 below. The adaptive interference alignment method in a time varying multi-user multi-antenna interference channel environment. [Equation 7]

The method according to claim 4, Using v [n] and H _ij [n] obtained in the first process, using Equation 10 below

To repeat the process of renewing

Adaptive interference alignment method in a time-varying multi-user multi-antenna interference channel environment characterized in that it comprises a second step of obtaining. [Equation 10]

The method according to claim 5, Obtained in the second step

And using the prediction channel matrix H _ij [n] of the current time,

Adaptive interference alignment method in a time-varying multi-user multi-antenna interference channel environment comprising a third step of configuring a. [Matrix]

The method according to claim 6,

Eigenvalues

And eigenvalue vectors Adaptive interference alignment method in a time-varying multi-user multi-antenna interference channel environment comprising the fourth step of obtaining a. The method of claim 7, Determine v [n] = q ₁ and use it as the beamforming vector at the current time n = kL, and the remaining eigenvectors and eigenvalues before the next period (n = kL + 1, ..., kL + L-1) Adaptive interference alignment method in a time-varying multi-user multi-antenna interference channel environment comprising a fifth process for storing the beamforming vector. The method according to claim 8, from n = kL

With vectors

And

Adaptive method in a time-varying multi-user multi-antenna interference channel environment comprising a sixth process of obtaining a beamforming vector without iterative calculation at time n = kL + 1, ..., kL + L-1 using Interference Alignment Method. delete

Images