CN106788642B - Hybrid precoding design method for actual broadband large-scale MIMO system - Google Patents

Abstract

The invention provides a hybrid precoding design method for an actual broadband large-scale MIMO system, which comprises the following steps: at a radio frequency end, firstly, assuming that analog precoding on all carriers is the same, and calculating an ideal analog precoding matrix by using complete channel state information by taking the maximum system spectrum efficiency as a criterion; then according to the characteristics of a phase shifter in an actual system, determining a simulated precoding matrix on each carrier in practice, designing a phase correction matrix in a digital domain, correcting the phase deviation of the simulated precoding on different carriers in practice to approximate to ideal simulated precoding, and multiplying the phase correction matrix and the simulated precoding matrix with the phase deviation on each carrier to obtain the designed simulated precoding matrix; and finally, designing a digital precoding matrix by using equivalent low-dimensional channel state information at a baseband, and multiplying the digital precoding matrix by the obtained analog precoding to obtain a mixed precoding design scheme.

Description

A Hybrid Precoding Design Method for Practical Wideband Massive MIMO Systems

Technical field:

The invention belongs to the technical field of wireless communication, and in particular relates to a hybrid precoding design method for an actual wideband massive MIMO system.

Background technique:

Hybrid precoding is a research hotspot for massive MIMO. When the number of antennas is large (hundreds or thousands), it is impossible to equip each antenna with a specific radio frequency link (RF) due to power consumption and cost. It becomes very interesting to investigate the deployment of massive MIMO with a small amount of RF. Hybrid precoding uses a low-cost phase shifter on the radio side to control the phase of the signal on the transmit antenna to realize analog precoding, which not only reduces the hardware cost, but also reduces the number of radio frequencies required by the system; at the baseband, it uses an equivalent low-dimensional Channel State Information (CSI) controls the amplitude and phase of the signal, enabling digital precoding. Therefore, hybrid precoding can realize massive MIMO when the number of RFs is much smaller than the number of antennas. At present, the research on hybrid precoding mainly focuses on single-carrier systems, and there are relatively few researches on hybrid precoding in wideband systems, and the current design of wideband hybrid precoding assumes that the analog precoding on each subcarrier is the same condition. This is only suitable when the ratio of bandwidth to center carrier frequency is relatively small. However, considering the application of millimeter waves (30-300 GHz) in the future, the above assumptions are often unrealistic in practice. This is because the phase shifter in the broadband beamforming network is usually implemented with a delay line, which causes the same delay to produce different phases on different carriers. That is to say, although we only set an analog precoding matrix, but It produces phase offsets on different carriers. Therefore, in practice, the analog precoding on different carriers is different, which will bring about a non-negligible performance loss.

In the current research on hybrid precoding design of wideband systems, the non-ideal characteristics of hardware implementation in practical systems are rarely considered; in the case of narrow bandwidths in the past, the phase shifter's phase change with frequency is not too large, Its effect can be ignored. The current millimeter wave technology is ultra-wideband. For example, in the 60GHz frequency band, the bandwidth is generally 2G. On the other hand, in such a high frequency band, in order to ensure the performance of the system, the processing accuracy of the hardware is very fine, resulting in the high price of millimeter wave devices, so an intermediate frequency link is required. The common intermediate frequency is 2.75GHz, which will lead to a ratio of bandwidth to carrier frequency close to 0.5 to 1.5, so the influence of phase offset on analog precoding cannot be ignored.

In summary, it is necessary to study the hybrid precoding design method suitable for practical wideband massive MIMO systems.

Invention content:

The purpose of the present invention is to solve the above problems, and propose a hybrid precoding design method for practical wideband massive MIMO systems, which can improve the hybrid precoding performance of practical wideband systems.

To achieve the above object, the present invention adopts the following technical solutions to realize:

A hybrid precoding design method for a practical wideband massive MIMO system, comprising the following steps:

1) At the radio frequency side, first assume that the analog precoding on all carriers is the same, and use the complete channel state information to design an ideal analog precoding matrix with the criterion of maximizing the system spectral efficiency;

2) According to the characteristics of the phase shifter in practice, determine the analog precoding matrix with phase offset on each carrier, design the phase correction matrix in the digital domain, and correct the phase offset generated by analog precoding on different carriers in practice, With the ideal analog precoding matrix obtained in approximation step 1), the phase correction matrix is multiplied by the analog precoding matrix that phase offset occurs on each carrier to be the designed analog precoding matrix;

3) At the baseband, use the equivalent low-dimensional channel state information to design a digital precoding matrix;

4) Multiply the analog precoding obtained in step 2) and the digital precoding obtained in step 3) to obtain a hybrid precoding design scheme.

A further improvement of the present invention is that the concrete realization method of step 1) is as follows:

Consider a downlink broadband massive MIMO system, the base station transmits N _s data streams through N _RF radio frequencies and N _t antennas, the user configures N _r antennas, the number of radio frequencies is the same as the number of antennas, and there are K sub-carriers in total; F _BB [k] represents the digital baseband precoding matrix on the kth carrier, F _RF represents the analog precoding matrix on all carriers, and at the receiving end, W[k] represents the receiving combining matrix, where N _t >>N _RF ;

101) Received signal on the kth carrier:

y[k]=W ^H [k]H[k]F _RF F _BB [k]s[k]+W ^H [k]n[k] (1)

表示第k个载波上的噪声；where H[k] represents the channel on the kth carrier,

represents the noise on the kth carrier;

102) According to the received signal in the previous step, the base station side designs analog precoding and digital precoding with the goal of maximizing mutual information and is described as follows:

是模拟预编码的可行集，即一组所有元素幅度都相同的N_t×N_RF矩阵集；in,

is a feasible set of analog precoding, that is, a set of N _t ×N _RF matrices with the same magnitude of all elements;

103) Solve the above optimization problem to obtain an ideal analog precoding matrix:

where R is the channel correlation matrix.

The further improvement of the present invention is, the concrete realization method of step 2) is as follows:

201) According to the characteristics of the phase shifter in practice, determine the analog precoding matrix with phase offset on each carrier: take the analog precoding matrix obtained in step 1) as the analog precoding matrix on the center carrier frequency, then in practice The analog precoding matrix on the kth carrier can be expressed as:

表示第k个载波上的相位偏差矩阵，

表示矩阵A与矩阵B的Hadamard积；where, denotes, F _RF is the analog precoding matrix on the center carrier frequency,

represents the phase deviation matrix on the kth carrier,

Represents the Hadamard product of matrix A and matrix B;

202) Correct the phase offset in the digital domain: the correction scheme can be described as:

可在步骤1)求得，当第k个载波的频率给定后，由公式(6)计算得F_RF[k]，因此，求解P2，得到一个众所周知最小均方解：in

It can be obtained in step 1). When the frequency of the kth carrier is given, F _RF [k] is calculated by formula (6). Therefore, solving P2, a well-known least mean square solution is obtained:

表示矩阵A的伪逆；in

represents the pseudo-inverse of matrix A;

203) Multiplying the phase correction matrix by the analog precoding matrix with phase offset on each carrier is the designed analog precoding matrix, that is, F _RF [k]F _c [k].

A further improvement of the present invention is that the concrete realization method of step 3) is as follows:

After determining the analog precoding and phase correction matrices on all carriers in step 2), the equivalent low-dimensional channel state information at the baseband is expressed as:

，其中V_eff[k]为等效信道H_eff[k]的SVD分解的右奇异矩阵。At baseband, digital precoding can be obtained by SVD decomposition of the equivalent low-dimensional channel state information

, where V _eff [k] is the right singular matrix of the SVD decomposition of the equivalent channel _{He eff} [k].

A further improvement of the present invention is that the concrete realization method of step 4) is as follows:

Multiply the analog precoding obtained in step 2) and the digital precoding obtained in step 3) to obtain the hybrid precoding design scheme:

F=F _RF [k] F _c [k] F _BB [k] (10).

Compared with the prior art, the present invention has the following advantages:

For the first time, the present invention considers the performance that the phase shifter in the broadband beamforming network is more sensitive to frequency changes, and designs a hybrid precoding scheme that is more in line with the actual system. Compared with the traditional hybrid precoding scheme, on the basis of designing an ideal analog precoding, the present invention proposes a method for the actual situation that the analog precoding on different carriers is different due to the phase offset generated by the phase shifter. It is an effective solution to set the phase correction matrix in the digital domain to correct the phase offset, thereby improving the overall spectral efficiency of the system and greatly improving the frequency efficiency of the edge carrier.

Description of drawings:

Fig. 1 is the spectral efficiency comparison diagram on each carrier corresponding to different schemes;

FIG. 2 is a comparison diagram of the average spectral efficiency on all carriers corresponding to different schemes.

Detailed ways:

Below in conjunction with accompanying drawing, the present invention is described in further detail:

The main idea of the hybrid precoding design method for an actual wideband massive MIMO system provided by the present invention is as follows: the first step, at the radio frequency side, assumes that the analog precoding on all carriers is the same, and uses the complete channel state information to design The ideal analog precoding matrix; in the second step, according to the characteristics of the phase shifter in practice, determine the analog precoding matrix with phase offset on each carrier, design the phase correction matrix in the digital domain, and correct the actual phase offset on different carriers. To approximate the ideal analog precoding matrix obtained in the first step, multiply the phase correction matrix by the analog precoding matrix with phase offset on each carrier to obtain the designed analog precoding. coding matrix; the third step, at the baseband, use the equivalent low-dimensional channel state information to design a digital precoding matrix; the fourth step, the analog precoding obtained in the second step and the digital precoding obtained in the third step Multiplying, that is, the hybrid precoding design scheme.

The specific implementation is as follows:

Consider a downlink wideband massive MIMO system, the base station transmits N _s data streams through N _RF radio frequencies and N _t (N _t >> N _RF ) antennas, the user configures N _r antennas, and the number of radio frequencies is the same as the number of antennas , with a total of K subcarriers. _FBB [k] represents the digital baseband precoding matrix on the kth carrier, _FRF represents the analog precoding matrix on all carriers, and at the receiving end, W[k] represents the receiving combining matrix;

On the basis of the above, the received signal on the kth carrier can be expressed as

y[k]=W ^H [k]H[k]F _RF F _BB [k]s[k]+W ^H [k]n[k] (1)

表示第k个载波上的噪声。where H[k] represents the channel on the kth carrier,

represents the noise on the kth carrier.

The spectral efficiency on each carrier and the total system spectral efficiency are shown in equation (2) and equation (3), respectively:

The technical means of the hybrid precoding design method suitable for the actual wideband massive MIMO system proposed by the present invention are as follows:

First, assuming that the analog precoding on all carriers is the same, the base station designs analog precoding and digital precoding with the goal of maximizing mutual information, which can be described as follows:

是模拟预编码的可行集，即一组所有元素幅度都相同的N_t×N_RF矩阵集。in,

is the feasible set of analog precoding, that is, a set of N _t ×N _RF matrices with the same magnitude of all elements.

Solving the above optimization problem yields an ideal analog precoding:

where R is the channel correlation matrix;

Then, the obtained ideal analog precoding matrix is used as the analog precoding matrix on the center carrier frequency. According to the characteristics of the phase shifter in practice, the analog precoding matrix on the kth carrier in practice can be expressed as:

表示第k个载波上的相位偏差矩阵，

表示矩阵A与矩阵B的Hadamard积。where, denotes, F _RF is the analog precoding matrix on the center carrier frequency,

represents the phase deviation matrix on the kth carrier,

Represents the Hadamard product of matrix A and matrix B.

In order to improve the performance of the system, the phase offset needs to be corrected in the digital domain. The correction scheme can be described as

为上述理想的模拟预编码矩阵，当第k个载波的频率给定后，可由公式(6)计算得F_RF[k]。因此，求解P2，我们可以得到一个众所周知最小均方解：in

For the above-mentioned ideal analog precoding matrix, when the frequency of the kth carrier is given, F _RF [k] can be calculated by formula (6). Therefore, solving for P2, we can obtain a well-known least-mean-square solution:

表示矩阵A的伪逆，将相位修正矩阵与每个载波上发生相位偏移的模拟预编码矩阵相乘即为所设计的模拟预编码矩阵，即F_RF[k]F_c[k]；in

Represents the pseudo-inverse of matrix A, and multiplying the phase correction matrix by the analog precoding matrix with phase offset on each carrier is the designed analog precoding matrix, namely F _RF [k]F _c [k];

On this basis, using the equivalent channel state information, the digital precoding matrix is designed. After the above steps, the analog precoding and phase correction matrices on all carriers are determined, and the equivalent low-dimensional channel state information at the baseband can be expressed as:

，其中V_eff[k]为等效信道H_eff[k]的SVD分解的右奇异矩阵；At baseband, digital precoding is obtained by SVD decomposition of the equivalent low-dimensional channel state information

, where V _eff [k] is the right singular matrix of the SVD decomposition of the equivalent channel H _eff [k];

Finally, multiply the analog precoding and digital precoding obtained in the above steps to obtain the hybrid precoding design scheme:

F=F _RF [k] F _c [k] F _BB [k] (10)

The simulation effect of the present invention is as follows:

The number of base station antennas N _t =64, the number of radio frequencies N _RF =8, the number of data streams N _s =4, the number of user antennas N _r =4, the number of subcarriers K = 4096, and the ratio of broadband to center carrier frequency is 0.5-1.5. The present invention is compared with three schemes of all-digital precoding, ideal hybrid precoding and actual hybrid precoding without phase correction, and the comparison results are shown in FIG. 1 and FIG. 2 .

Figure 1 shows the spectral efficiency on each carrier corresponding to different schemes when the signal-to-noise ratio SNR=10dB. It can be seen that the spectral efficiency of the edge carrier of the uncorrected phase scheme is 2.2dB lower than that of the ideal scheme, while the phase correction scheme proposed in the present invention is 1.2dB higher than the uncorrected phase scheme, and is close to the ideal hybrid pre-processing to a certain extent. encoding scheme.

Figure 2 shows the variation of the average spectral efficiency on all carriers corresponding to different schemes with the signal-to-noise ratio. It can be seen from the figure that the overall performance of the uncorrected phase scheme system is 0.5dB lower than the ideal case, and the phase correction algorithm proposed in the present invention can eliminate the performance loss caused by the phase offset to a certain extent. The performance is about 0.3dB higher than the uncorrected phase scheme.

Claims (2)

1. a hybrid precoding design method for an actual wideband massive MIMO system, is characterized in that, comprises the following steps: 1) On the radio side, first assume that the analog precoding on all carriers is the same, and use the complete channel state information to design an ideal analog precoding matrix based on maximizing the system spectral efficiency. The specific implementation method is as follows: Consider a downlink broadband massive MIMO system, the base station transmits N _s data streams through N _RF radio frequencies and N _t antennas, the user configures N _r antennas, the number of radio frequencies is the same as the number of antennas, and there are K sub-carriers in total; F _BB [k] represents the digital baseband precoding matrix on the kth carrier, F _RF represents the analog precoding matrix on all carriers, and at the receiving end, W[k] represents the receiving combining matrix, where N _t >>N _RF ; 101) Received signal on the kth carrier: y[k]=W ^H [k]H[k]F _RF F _BB [k]s[k]+W ^H [k]n[k] (1) where H[k] represents the channel on the kth carrier,

represents the noise on the kth carrier; 102) According to the received signal in the previous step, the base station side designs analog precoding and digital precoding with the goal of maximizing mutual information and is described as follows:

in,

is a feasible set of analog precoding, a set of N _t ×N _RF matrices with the same magnitude of all elements; 103) Solve the above optimization problem to obtain an ideal analog precoding matrix:

where R is the channel correlation matrix; 2) According to the characteristics of the phase shifter in practice, determine the analog precoding matrix with phase offset on each carrier, design the phase correction matrix in the digital domain, and correct the phase offset generated by analog precoding on different carriers in practice, By approximating the ideal analog precoding matrix obtained in step 1), multiplying the phase correction matrix and the analog precoding matrix with phase offset on each carrier is the designed analog precoding matrix; The specific implementation method is as follows: 201) According to the characteristics of the phase shifter in practice, determine the analog precoding matrix with phase offset on each carrier: take the analog precoding matrix obtained in step 1) as the analog precoding matrix on the center carrier frequency, then in practice The analog precoding matrix on the kth carrier is expressed as:

where F _RF is the analog precoding matrix on the center carrier frequency,

represents the phase deviation matrix on the kth carrier,

Represents the Hadamard product of matrix A and matrix B; 202) Correct the phase offset in the digital domain: the correction scheme can be described as:

in

Obtained in step 1), when the frequency of the kth carrier is given, F _RF [k] is calculated by formula (6). Therefore, solve P2 to obtain a least mean square solution:

in

represents the pseudo-inverse of matrix A; 203) multiplying the phase correction matrix by the analog precoding matrix with phase offset on each carrier is the designed analog precoding matrix, namely F _RF [k]F _c [k]; 3) At the baseband, use the equivalent low-dimensional channel state information to design a digital precoding matrix; the specific implementation method is as follows: After determining the analog precoding and phase correction matrices on all carriers in step 2), the equivalent low-dimensional channel state information at the baseband is expressed as:

At baseband, digital precoding is obtained by SVD decomposition of the equivalent low-dimensional channel state information

where V _eff [k] is the right singular matrix of the SVD decomposition of the equivalent channel H _eff [k]; 4) Multiply the analog precoding obtained in step 2) and the digital precoding obtained in step 3) to obtain a hybrid precoding design scheme. 2. a kind of hybrid precoding design method for actual wideband massive MIMO system according to claim 1, is characterized in that, the concrete realization method of step 4) is as follows: Multiply the analog precoding obtained in step 2) and the digital precoding obtained in step 3) to obtain the hybrid precoding design scheme: F=F _RF [k] F _c [k] F _BB [k] (10).

Images