CN108600125B - An Iterative Channel Estimation Method - Google Patents

Abstract

The invention discloses an iterative-based channel estimation method used in a large-scale antenna system, comprising the following steps: step 1, designing equations for sending and receiving signals, including creating a training matrix and a received signal corresponding to the training matrix; step 2, Based on the iterative channel estimation, the channel is re-estimated using the equation of step one. Compared with the existing channel estimation based on pilot sequence, the channel estimation method proposed by the present invention improves the channel estimation accuracy, and its complexity is slightly higher, but the present invention has excellent channel estimation accuracy and greatly improves the channel estimation accuracy. system performance.

Description

An Iterative Channel Estimation Method

technical field

The invention belongs to the technical field of wireless communication, and in particular relates to a multi-cell multi-user large-scale antenna technology for wireless communication, in particular to the application of an iterative-based channel estimation method in a large-scale antenna system.

Background technique

MIMO (Multiple-Input Multiple-Output) technology refers to the use of multiple transmitting and receiving antennas at the transmitting and receiving ends, respectively, so that signals are transmitted and received through multiple antennas at the transmitting and receiving ends, thereby improving communication quality. A variety of technologies in multiple places. This technology can make full use of space resources, realize multiple transmission and multiple reception through multiple antennas, and can double the system channel capacity without increasing spectrum resources and antenna transmission power.

With the rapid development of wireless communication technology, traditional MIMO technology can no longer meet people's requirements for transmitting wireless data. Therefore, multi-user multiple-input multiple-output (MU-MIMO) wireless communication systems are widely used. MU-MIMO is the abbreviation of "Multi-User Multiple-Input Multiple-Output". It is a multi-user multiple-input multiple-output technology that allows the user's router to communicate with multiple devices at the same time. Among them, uplink MU-MIMO: Different users use the same time-frequency resources for uplink transmission (single-antenna transmission). From the receiving end, these data streams can be regarded as different antennas from a user terminal, thus forming a virtual MIMO system, that is, uplink MU- MIMO; downlink MU-MIMO: multiple data streams are transmitted to different user terminals, and multiple user terminals and eNB form a downlink MU-MIMO system; downlink MU-MIMO can separate transmissions at the receiving end by eliminating/nulling Data streams for different users; Downlink MU-MIMO can also separate data streams of different users in advance by adopting beamforming at the transmitting end, thereby simplifying the operation of the receiving end. This technology can ensure the rate and accuracy of wireless data transmission while multiple users in multiple cells communicate at the same time. However, compared with the traditional MIMO system, the MU-MIMO wireless communication system still has defects such as the randomness of the channel state information and the interference of users between cells.

In an uplink multi-cell multi-user MIMO system, in general, the base station performs channel estimation by sending respective training sequences from different users in the cell, and there will be severe mutual interference between training and information sequences in channel estimation and detection, making channel estimation and System performance degrades. Therefore, how to improve the accuracy of channel estimation through better estimation methods is of great significance to the improvement of system performance.

SUMMARY OF THE INVENTION

Aiming at the insufficient accuracy of the existing channel estimation, the present invention proposes an iterative-based channel estimation method, which is used in an uplink multi-cell multi-user large-scale antenna system. In the present invention, the receiving end obtains the estimated channel through the received pilot sequence, decodes it with minimum mean square error (MMSE), selects the qualified signal from the decoded signal and adds it to the original pilot sequence to form a new The pilot sequence is used to re-estimate the channel.

The present invention adopts following technical scheme:

An iterative-based channel estimation method, comprising the following steps:

Step 1, designing the equations of sending and receiving signals, including creating a training matrix and a received signal corresponding to the training matrix;

Step 2, based on the iterative channel estimation, re-estimate the channel using the equation of step 1.

Further, in the step 1, creating a training matrix includes:

1.1 Setting the training sequence, it is assumed that the system has L cells, each cell has one base station and K users, each base station is equipped with M antennas, and there is co-channel interference between cells. It is assumed that the kth user of the lth cell first sends a training sequence of length τ Φ _lk =(φ _lk,1 ,φ _lk,2 ,...,φ _lk,τ ), where φ _lk,1 ,φ _lk,2 ,...,φ _lk,τ , k=1,2,...,K is the training symbol sent by the kth user of the lth cell at the 1st, 2nd,...,τ moment respectively; l=1,2,. ..,L; L≥1, K≥1, M≥1, and take a natural number (integer), τ>0.

为K×τ矩阵，则Φ＝(Φ₁,Φ₂,…,Φ_L)^h为LK×τ的训练矩阵，其中上标h表示向量或矩阵的转置。1.2 Obtain the training matrix of LK×τ, let

is a K×τ matrix, then Φ=(Φ ₁ ,Φ ₂ ,…,Φ _L ) ^h is the training matrix of LK×τ, where the superscript h represents the transpose of the vector or matrix.

Further, in the step 1, the creation of the received signal corresponding to the training matrix includes:

表示第t时刻发送端KL个用户发送的信号，0≤t≤T，T＞0；1.3 Create a signal matrix X, assuming that the coherence time of the channel is T, within the coherence time, each user at the transmitter sends T pieces of data information. Then, within the coherence time, the sender sends a KL×T-dimensional signal matrix X, that is, X=(X ₁ , X ₂ ,...,X _T ); where,

Indicates the signal sent by KL users at the transmitting end at time t, 0≤t≤T, T>0;

1.4 The channel environment remains unchanged, assuming that the received signals corresponding to Φ and X in the first base station are:

Among them, H ₌ [H ₁ , H _{2 , .} [B ₁ ,B ₂ ,...,B _L ], where B _l is a K×K diagonal matrix, representing the large-scale attenuation factor from the user in the lth base station to the first base station antenna; Y=(Y ₁ , Y ₂ ,...,Y _T ) represents an M×T-dimensional matrix, and Y _t represents an M×1-dimensional signal vector received by the base station at time t, where t=1,2,...,T; W ₀ and Both W ₁ represent white Gaussian noise, and both ρ ₀ and ρ ₁ are signal-to-noise ratios.

Further, the LK×τ training matrix Φ=[I _k I _k ... I _k ] ^h , Φ includes L I _k , and each I _k represents a K-order identity matrix.

Further, the step 2 includes:

2.1 Estimate the channel for the first time, according to the transmitted training matrix Φ and the corresponding received signal Y ₀ , use the MMSE estimation method to estimate the channel H;

可得到解码滤波矩阵；2.2 Use the MMSE decoding method to decode the received signal at the receiving end, and estimate the channel by

The decoding filter matrix can be obtained;

2.3 Decoding signal screening, screening out the signal matrix of L-1 cells before m times, where m>0, to obtain the received signal matrix;

2.4 Re-estimate the channel, the MMSE criterion estimates the channel to obtain a new channel matrix,

2.5 Iterative channel-based MMSE decoding with re-estimated channel.

Further, the step 2.1 is specifically:

According to the transmitted training matrix Φ and the corresponding received signal Y ₀ , the channel H is estimated by the MMSE estimation method, and the estimated channel can be obtained:

为M×LK的矩阵，令

其中

为M×K的矩阵。in

is an M×LK matrix, let

in

is an M×K matrix.

Further, the step 2.2 is specifically:

可得到解码滤波矩阵为：The received signal is decoded at the receiving end using the MMSE decoding method, and the channel is estimated by

The decoding filter matrix can be obtained as:

σ²＝1+ρ₁tr(BA^-1B)。在相干时间T内，接收端通过MMSE准则可以得到T组前L-1个小区的解码信号。in:

σ ² =1+ρ ₁ tr(BA ⁻¹ B). Within the coherence time T, the receiving end can obtain the decoded signals of the first L-1 cells in the T group through the MMSE criterion.

中减去前L-1个小区的估计接收信号得到第L个小区的估计接收信号

可以表示为：Further, the step 2.3 is specifically: using the screening criteria of the decoded signal to screen the signal, after the screening of the signal, it is assumed that the signal matrix of the L-1 cells before m times can be screened, and the first L-1 cells can be screened out. The signal of the cell is used to estimate the signal of the Lth cell. At time t _i , when the signal is received

The estimated received signal of the L-th cell is obtained by subtracting the estimated received signal of the first L-1 cells from

It can be expressed as:

Then the estimated Lth cell transmits signal information:

及相对应的M×m维接收信号矩阵

记：The KL×m-dimensional signal matrix can be obtained

and the corresponding M×m-dimensional received signal matrix

remember:

x_l,i＝(x_l1,i,x_l2,i,…,x_lK,i)^h，i＝t₁,t₂,…,t_m，l＝1,2,…,L。in,

xl _,i =( _xl1,i , _xl2,i ,..., _xlK,i ) ^h , i= _t1 , _t2 ,..., _tm , l=1,2,...,L.

Further, in the screening of the decoded signal in the described step 2.3, the screening criterion utilized is

Criterion 1, the user signals of different cells at the transmitter are taken from the same constellation QPSK,

And at time t,

or,

Criterion 2, the user signals of adjacent cells at the transmitting end are respectively taken from the constellation QPSK, QPSK ₁ ,

QPSK₁＝exp(jπ/4)×QPSK，

QPSK ₁ =exp(jπ/4)×QPSK,

中选出s列和矩阵Φ组成新的训练序列矩阵F，使矩阵F秩为最大，并将对应s个时刻的接收信号矩阵和矩阵Y₀组成新的接收矩阵Y₀₀。重新用MMSE准则估计信道得到新的信道矩阵：Further, the step 2.4 is specifically: from the selected m time signal matrices

Columns s and matrix Φ are selected to form a new training sequence matrix F, so that the rank of matrix F is the largest, and the received signal matrix and matrix Y ₀ corresponding to s times are formed into a new receiving matrix Y ₀₀ . Reuse the MMSE criterion to estimate the channel to obtain a new channel matrix:

可得MMSE解码滤波矩阵为：estimated channel by

The available MMSE decoding filter matrix is:

和新的解码滤波矩阵G_dmmse进行解码，则第一个小区的第k个用户的MMSE解码可以表示为：Further, the step 2.5 is specifically: using the newly estimated channel matrix

Decoding with the new decoding filter matrix G _dmmse , the MMSE decoding of the kth user in the first cell can be expressed as:

in,

Beneficial effects of the present invention: Compared with the existing channel estimation based on the pilot sequence, the channel estimation method proposed by the present invention improves the channel estimation accuracy, and its complexity is slightly higher, but the present invention has excellent channel estimation accuracy , and greatly improve the system performance.

Description of drawings

1 is a simulation diagram about channel estimation error under the conditions of Embodiment 2;

FIG. 2 is a simulation diagram of channel estimation performance under the conditions of Embodiment 2. FIG.

Detailed ways

The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings to make the solution more clear.

Example 1

This embodiment discloses an iterative-based channel estimation method applied to a multi-cell multi-user large-scale antenna involving wireless communication, the method specifically includes the following steps:

1. Design transmit and receive signal equations

为K×τ矩阵，Φ＝(Φ₁,Φ₂,…,Φ_L)^h为LK×τ的训练矩阵，上标h表示向量或矩阵的转置(以下同)。然后假设信道的相干时间为T，在相干时间内，发送端每个用户发送T个数据信息。则相干时间内，发送端发送KL×T维的信号矩阵X，即X＝(X₁,X₂,…,X_T)。其中，

表示第t时刻，发送端KL个用户发送的信号。由于信道环境不变，假设第1个基站中与Φ和X对应的接收信号分别为：It is assumed that the system has L cells, each cell has one base station and K users, each base station is equipped with M antennas, and there is co-channel interference between cells. It is assumed that the kth user of the lth cell first sends a training sequence of length τ, that is, Φ _lk =(φ _lk,1 ,φ _lk,2 ,...,φ _lk,τ ), where φ _lk,1 ,φ _{lk ,2} ,...,φ _lk,τ , k=1,2,...,K is the training symbol sent by the kth user of the lth cell at the 1st, 2nd,...,τ time instants, respectively. make

is a K×τ matrix, Φ=(Φ ₁ ,Φ ₂ ,…,Φ _L ) ^h is the training matrix of LK×τ, and the superscript h represents the transpose of the vector or matrix (the same below). Then, assuming that the coherence time of the channel is T, within the coherence time, each user at the transmitter sends T pieces of data information. Then, within the coherence time, the transmitting end sends a KL×T-dimensional signal matrix X, that is, X=(X ₁ , X ₂ , . . . , X _T ). in,

Indicates the signals sent by KL users at the sender at the t-th time. Since the channel environment remains unchanged, it is assumed that the received signals corresponding to Φ and X in the first base station are:

Wherein, H ₌ [H ₁ , H _{2 , .} B=diag[B ₁ , B ₂ , . . . , B _L ], where B _l is a K×K diagonal matrix, which represents the large-scale attenuation factor from the user in the lth base station to the first base station antenna. Y=(Y ₁ , Y ₂ ,...,Y _T ) represents an M×T-dimensional matrix, and Y _t represents an M×1-dimensional signal vector received by the base station at the t-th time, where t=1,2,..., T. In the present invention, let Φ=[I _k I _k ... I _k ] ^h , Φ includes L I _k s, and each I _k represents a K-order unit matrix (the same below). Both W ₀ and W ₁ represent white Gaussian noise. Both ρ ₀ and ρ ₁ are signal-to-noise ratios.

In order to improve the accuracy of channel estimation, the present invention proposes an iteration-based channel estimation method by screening the decoded signal and adding it to the training sequence. Specifically as follows:

2. Iterative-based channel estimation, including the following 5 steps,

1. Estimate the channel for the first time

According to the transmitted training matrix Φ and the corresponding received signal Y ₀ , the channel H is estimated by the MMSE estimation method, and the estimated channel can be obtained:

为M×LK的矩阵。令

其中

为M×K的矩阵。in

is an M×LK matrix. make

in

is an M×K matrix.

2. Receiver decoding

可得到解码滤波矩阵为：The received signal is decoded at the receiving end using the MMSE decoding method, and the channel is estimated by

The decoding filter matrix can be obtained as:

σ²＝1+ρ₁tr(BA^-1B)。在相干时间T内，接收端通过MMSE准则可以得到T组前L-1个小区的解码信号。in:

σ ² =1+ρ ₁ tr(BA ⁻¹ B). Within the coherence time T, the receiving end can obtain the decoded signals of the first L-1 cells in the T group through the MMSE criterion.

3. Decoding Signal Screening

In the present invention, two methods for selecting the constellation diagram of the transmitting end will be proposed. In the first method, user signals of different cells at the transmitter are taken from the same constellation QPSK. In the second method, the user signals of adjacent cells at the transmitting end are respectively obtained from the constellation QPSK and QPSK ₁ . in:

QPSK₁＝exp(jπ/4)×QPSK

QPSK ₁ =exp(jπ/4)×QPSK

And at time t, the user signal screening criterion in method 1 is:

The screening criteria for the decoded signal in Method 2 are:

中减去前L-1个小区的估计接收信号得到第L个小区的估计接收信号

可以表示为：After signal screening, it is assumed that the signal matrix of L-1 cells before m times can be screened out, and the signals of the first L-1 cells are used to estimate the signal of the Lth cell. At time t _i , when the signal is received

The estimated received signal of the L-th cell is obtained by subtracting the estimated received signal of the first L-1 cells from

It can be expressed as:

Then the estimated Lth cell transmits signal information:

及相对应的M×m维接收信号矩阵

记：

其中，Therefore, the KL×m-dimensional signal matrix can be obtained

and the corresponding M×m-dimensional received signal matrix

remember:

in,

x_l,i＝(x_l1,i,x_l2,i,…,x_lK,i)^h

x _l,i =(x _l1,i ,x _l2,i ,…,x _lK,i ) ^h

i=t ₁ , t ₂ ,...,t _m , l=1,2,...,L.

4. Re-estimate the channel

中选出s列和矩阵Φ组成新的训练序列矩阵F，使矩阵F秩为最大，并将对应s个时刻的接收信号矩阵和矩阵Y₀组成新的接收矩阵Y₀₀。重新用MMSE准则估计信道得到新的信道矩阵：From the selected m time signal matrix

Columns s and matrix Φ are selected to form a new training sequence matrix F, so that the rank of matrix F is the largest, and the received signal matrix and matrix Y ₀ corresponding to s times are formed into a new receiving matrix Y ₀₀ . Reuse the MMSE criterion to estimate the channel to obtain a new channel matrix:

可得MMSE解码滤波矩阵为：estimated channel by

The available MMSE decoding filter matrix is:

5. MMSE decoding based on iterative channel

和新的解码滤波矩阵G_dmmse进行解码，则第一个小区的第k个用户的MMSE解码数据

可以表示为：with the newly estimated channel matrix

Decode with the new decoding filter matrix G _dmmse , then the MMSE decoded data of the kth user of the first cell

It can be expressed as:

in,

Example 2

Referring to FIG. 1, it is a simulation diagram of a comparison of channel estimation errors under different channel estimations when M=128, and FIG. 2 is a simulation diagram of a comparison of average bit error rates under M=128 and different channel estimations.

This embodiment discloses an iterative-based channel estimation method applied to a multi-cell multi-user large-scale antenna involving wireless communication, the method specifically includes the following steps:

1. Design transmit and receive signal equations

为3×3矩阵，Φ＝(Φ₁,Φ₂,Φ₃)^h为9×3的训练矩阵，上标t表示向量或矩阵的转置(以下同)。然后假设信道的相干时间为T＝128，在相干时间内，发送端每个用户发送128个数据信息。则相干时间内，发送端发送9×128维的信号矩阵X，即X＝(X₁,X₂,…,X₁₂₈)。其中，

表示第t时刻，发送端9个用户发送的信号。由于信道环境不变，假设第1个基站中与Φ和X对应的接收信号分别为：It is assumed that the system has 3 cells, each cell has one base station and 3 users, each base station is equipped with 128 antennas, and there is co-channel interference between cells. Assuming that the kth user of the lth cell first sends a training sequence of length τ=3, that is, Φ _lk =(φ _lk,1 ,φ _lk,2 ,φ _lk,3 ), let

is a 3×3 matrix, Φ=(Φ ₁ ,Φ ₂ ,Φ ₃ ) ^h is a 9×3 training matrix, and the superscript t represents the transpose of the vector or matrix (the same below). Then, assuming that the coherence time of the channel is T=128, within the coherence time, each user at the transmitter sends 128 pieces of data information. Then, within the coherence time, the transmitting end sends a 9×128-dimensional signal matrix X, that is, X=(X ₁ , X ₂ , . . . , X ₁₂₈ ). in,

Indicates the signals sent by 9 users at the transmitter at the t-th time. Since the channel environment remains unchanged, it is assumed that the received signals corresponding to Φ and X in the first base station are:

Wherein, H=(H ₁ , H ₂ , H ₃ ), and H ₁ , H ₂ , and H ₃ are all 128×3 matrices. B=diag(B ₁ , B ₂ , B ₃ ), B ₁ , B ₂ , B ₃ are all 3×3 diagonal matrices, Y=(Y ₁ , Y ₂ ,...,Y ₁₂₈ ) means 128×128 dimensional matrix, Y _t represents the 128×1-dimensional signal vector received by the base station at time t. In the present invention, let Φ=(I ₃ , I ₃ , I ₃ ) ^h . Both W ₀ and W ₁ represent white Gaussian noise. Both ρ ₀ and ρ ₁ are signal-to-noise ratios.

2. Iterative-based channel estimation

1. Estimate the channel

According to the transmitted training matrix Φ and the corresponding received signal Y ₀ , the channel H is estimated by the MMSE estimation method, and the estimated channel can be obtained

为128×9的矩阵。令

其中

为128×3的矩阵。in

is a 128×9 matrix. make

in

is a 128x3 matrix.

2. Receiver decoding

可得到解码滤波矩阵为：The received signal is decoded at the receiving end using the MMSE decoding method, and the channel is estimated by

The decoding filter matrix can be obtained as:

Then, at time t, the MMSE decoding of the kth user in the first cell can be expressed as follows:

And the MMSE decoding of the kth user in the second cell can be expressed as:

Within the coherence time T, the receiver can obtain 128 sets of decoded signals of the first two cells through the MMSE criterion.

3. Decoding Signal Screening

In the present invention, two methods for selecting the constellation diagram of the transmitting end will be proposed. In the first method, user signals of different cells at the transmitter are taken from the same constellation QPSK. In the second method, the user signals of adjacent cells at the transmitting end are respectively obtained from the constellation QPSK and QPSK ₁ . in:

QPSK₁＝exp(jπ/4)×QPSK。

QPSK ₁ =exp(jπ/4)×QPSK.

And at time t, the user signal screening criterion in method 1 is:

That is, the signals decoded by the kth user in the first two cells are the same. The screening criteria for the decoded signal in Method 2 are:

中减去前2个小区的估计接收信号得到第3个小区的估计接收信号

可以表示为：After signal screening, it is assumed that the signal matrix of the first two cells at m time can be screened out, and the signals of the first two cells are used to estimate the signal of the third cell. At time t _i , when the signal is received

Subtract the estimated received signal of the first 2 cells from the estimated received signal of the third cell to obtain the estimated received signal of the third cell

It can be expressed as:

Then the estimated third cell sends signal information:

及相对应的M×m维接收信号矩阵

记：Therefore, a KL×m-dimensional signal matrix can be obtained

and the corresponding M×m-dimensional received signal matrix

remember:

其中，

in,

x_l,i＝(x_l1,i,x_l2,i,…,x_lK,i)^h

x _l,i =(x _l1,i ,x _l2,i ,…,x _lK,i ) ^h

i=t ₁ , t ₂ ,...,t _m , l=1,2,...,L.

4. Iterative-based channel estimation

中选出s列和矩阵Φ组成新的训练序列矩阵F，使矩阵F秩为最大，并将对应s个时刻的接收信号矩阵和矩阵Y₀组成新的接收矩阵Y₀₀。重新用MMSE准则估计信道得到新的信道矩阵：From the selected m time signal matrix

Columns s and matrix Φ are selected to form a new training sequence matrix F, so that the rank of matrix F is the largest, and the received signal matrix and matrix Y ₀ corresponding to s times are formed into a new receiving matrix Y ₀₀ . Reuse the MMSE criterion to estimate the channel to obtain a new channel matrix:

可得MMSE解码滤波矩阵为：estimated channel by

The available MMSE decoding filter matrix is:

5. MMSE decoding based on iterative channel

和新的解码滤波矩阵G_dmmse进行解码，则第一个小区的第k个用户的MMSE解码可以表示为：with the newly estimated channel matrix

Decoding with the new decoding filter matrix G _dmmse , the MMSE decoding of the kth user in the first cell can be expressed as:

in,

As shown in FIG. 1 , it is a simulation diagram of channel estimation error under the conditions of the above-mentioned second embodiment. As can be seen from FIG. 1 , compared with the traditional channel estimation method based on training sequence, the channel estimation method of the present invention improves the channel estimation accuracy and reduces the channel estimation error. However, method 2 can make the new training sequence matrix F reach full rank after iteration, thereby further improving the accuracy of channel estimation. As shown in FIG. 2, it is a simulation diagram about the system performance under the conditions of the above-mentioned Embodiment 1, and the system performance is based on the average decoding error probability of users in the target cell. It can be seen from Fig. 2 that the channel estimation method in the present invention improves the accuracy of channel estimation, thus improves the system performance and reduces the decoding bit error rate. However, under method 2, due to the pilot sequence in the iterative estimation process When the full rank is reached, the system performance has been greatly improved.

Those of ordinary skill in the art should realize that the above examples are only used to illustrate the present invention, not as a limitation of the present invention. As long as the changes and deformations of the above examples are within the scope of the present invention, all changes and modifications of the above examples will fall within the protection of the present invention. scope.

Claims (9)

1. an iterative-based channel estimation method, characterized in that, comprising the steps: Step 1, designing the equations of sending and receiving signals, including creating a training matrix and a received signal corresponding to the training matrix; Step 2, based on the iterative channel estimation, re-estimate the channel by using the equation of step 1; the step 2 includes: 2.1 Estimate the channel for the first time, according to the transmitted training matrix Φ and the corresponding received signal Y ₀ , use the MMSE estimation method to estimate the channel H; 2.2 Use the MMSE decoding method to decode the received signal at the receiving end, and estimate the channel by

The decoding filter matrix can be obtained; 2.3 Decoding signal screening, screening out the signal matrix of L-1 cells before m times, where m>0, to obtain the received signal matrix; 2.4 Re-estimate the channel, the MMSE criterion estimates the channel to obtain a new channel matrix, 2.5 Iterative channel-based MMSE decoding with re-estimated channel. 2. An iterative-based channel estimation method according to claim 1, wherein in the step 1, creating a training matrix comprises: 1.1 Setting the training sequence, it is assumed that the system has L cells, each cell has one base station and K users, each base station is equipped with M antennas, and there is co-channel interference between cells; Each user first sends a training sequence of length τ Φ _lk =(φ _lk,1 ,φ _lk,2 ,...,φ _lk,τ ), where φ _lk,1 ,φ _lk,2 ,...,φ _lk,τ , k=1,2,...,K is the training symbol sent by the kth user of the lth cell at the 1st, 2nd,...,τ moment respectively; L≥1, K≥1, M≥1, L, K , M is a natural number, τ>0; 1.2 Obtain the training matrix of LK×τ, let

is a K×τ matrix, then Φ=(Φ ₁ ,Φ ₂ ,…,Φ _L ) ^h is the training matrix of LK×τ, where the superscript h represents the transpose of the vector or matrix. 3. The iterative-based channel estimation method according to claim 2, wherein in the step 1, the creation of the received signal corresponding to the training matrix comprises: 1.3 Create a signal matrix X, assuming that the coherence time of the channel is T, within the coherence time, each user at the transmitter sends T pieces of data information. Then, within the coherence time, the sender sends a KL×T-dimensional signal matrix X, that is, X=(X ₁ , X ₂ ,...,X _T ), where,

Indicates the signal sent by KL users at the transmitting end at time t, 0≤t≤T, T>0; 1.4 The channel environment remains unchanged, assuming that the received signals corresponding to Φ and X in the first base station are:

Among them, H=(H ₁ , H ₂ ,...,H _L ), H _l represents the channel gain from the user in the lth base station to the first base station antenna, obviously H _l is an M×K matrix; B=diag ( _{B 1} , _{B 2} , _. , Y ₂ ,...,Y _T ) represents an M×T-dimensional matrix, and Y _t represents an M×1-dimensional signal vector received by the base station at time t, where t=1,2,...,T; W ₀ and Both W ₁ represent white Gaussian noise, and both ρ ₀ and ρ ₁ are signal-to-noise ratios. 4. An iterative-based channel estimation method according to claim 3, wherein the step 2.1 is specifically: According to the transmitted training matrix Φ and the corresponding received signal Y ₀ , the channel H is estimated by the MMSE estimation method, and the estimated channel can be obtained:

in,

is an M×LK matrix, let

in,

is an M×K matrix. 5. An iterative-based channel estimation method according to claim 4, wherein the step 2.2 is specifically: The received signal is decoded at the receiving end using the MMSE decoding method, and the channel is estimated by

The decoding filter matrix can be obtained as:

in:

Within the coherence time T, the receiving end can obtain the decoded signals of the first L-1 cells in the T group through the MMSE criterion. 6. An iterative-based channel estimation method according to claim 5, characterized in that, the step 2.3 is specifically: using the screening criterion of the decoded signal to screen the signal, after the screening of the signal, it is assumed that m can be screened out The signal matrix of the L-1 cells before the time, and the signal of the first L-1 cells is used to estimate the signal of the Lth cell. At time t _i , when the signal is received

The estimated received signal of the L-th cell is obtained by subtracting the estimated received signal of the first L-1 cells from , which can be expressed as:

Then the estimated Lth cell transmits signal information:

The KL×m-dimensional signal matrix can be obtained

and the corresponding M×m-dimensional received signal matrix

remember:

in,

i=t ₁ , t ₂ , ..., t _m , l=1, 2, ..., L. 7. The iterative-based channel estimation method according to claim 6, wherein the step 2.3 is specifically: Criterion 1, the user signals of different cells at the transmitter are taken from the same constellation QPSK,

And at time t,

k=1,2,...,K; Or, according to the second criterion, the user signals of adjacent cells at the transmitting end are respectively obtained from the constellation QPSK, QPSK ₁ ,

QPSK ₁ =exp(jπ/4)×QPSK,

l=1,2,...,L-1,k=1,2,...,K. 8. An iterative-based channel estimation method according to claim 7, characterized in that, the step 2.4 is specifically: from the selected m time signal matrices

Select S columns and matrix Φ to form a new training sequence matrix F, make the matrix F rank be the largest, and form a new receiving matrix Y ₀₀ corresponding to the received signal matrix and matrix Y ₀ of S moments; use the MMSE criterion again to estimate channel to get the new channel matrix:

estimated channel by

The available MMSE decoding filter matrix is:

9. The iterative-based channel estimation method according to claim 8, wherein the step 2.5 is specifically: using the newly estimated channel matrix

Decoding with the new decoding filter matrix G _dmmse , the MMSE decoding of the kth user in the first cell can be expressed as:

in,

Images