Precoding matrix selection method suitable for Type II

Abstract

The invention relates to the technical field of communication, in particular to a precoding matrix selection method suitable for TypeII, and the method comprises the following steps: S1, calculating the power Pbeam (i) of N1O1N2O2 beams, 0 < = ilt; n < 1 > O < 1 > N < 2 > O < 2 >; s2, comparing the signal-to-noise ratio of the CSI-RS signal with a preset threshold value to judge the channel quality of the current CSI-RS; s3, an appropriate beam scheme is selected according to the judgment result of the step S2 to calculate and obtain an optimal candidate beam W1, when the signal-to-noise ratio of the CSI-RS signal is smaller than a preset threshold value, it is indicated that the CSI-RS channel quality is poor, large errors exist in CSI-RS channel estimation, and a main beam selection scheme is adopted; when the signal-to-noise ratio of the CSI-RS signal is greater than the preset threshold value, the CSI-RS channel quality is good, the CSI-RS channel estimation is accurate, and a scheme of selecting a plurality of candidate beam combinations is adopted. According to the invention, different beam selection schemes are executed according to different channel qualities, the realization complexity can be greatly reduced when the performance loss is small, and the system performance is effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and more specifically, to a precoding matrix selection method applicable to Type II. Background Art

[0002] Since 4G LTE, multiple-input multiple-output transmission based on large-scale antenna arrays has been one of the effective means to continuously improve spectrum efficiency. By increasing the number of antenna units, the transmitted signal can be transmitted in a directional manner, thereby improving data transmission efficiency, reducing interference, and increasing the spectrum efficiency of the system. In 5G NR, data can be transmitted on a higher spectrum and there will be more antenna arrays. In order to more fully tap the potential of multiple antennas to improve spectrum efficiency, the standard designs different MIMO transmission precoding matrix feedback schemes to match the antenna characteristics and performance enhancement of multiple antennas.

[0003] In the NR system, all codebook designs adopt a parameterized codebook structure, using a two-level codebook W = W 1 · 2 , where W 1 Describes the long-term broadband characteristics of the channel, W 2 Describes the short-term subband characteristics of the channel. NR supports two types of codebooks for CSI feedback. One is the TypeI codebook with normal precision, which is mainly used for single-user transmission and supports ranks 1 to 8. The terminal needs to feedback the corresponding beam index and beam correlation index, so the feedback overhead is small. The other is the high-precision TypeII codebook, which can be used for single-user and multi-user transmission and supports ranks 1 to 4. It is mainly suitable for multi-user transmission and is equivalent to feeding back the right singular vector of the channel to the base station. The feedback overhead is large, but the performance gain is good. For TypeII, the basic principle is to transform the spatial channel into the angle domain. The merging coefficients represent the amplitude and phase of each beam component. On the one hand, the channel itself is sparse in space, with energy only in a few directions and zero energy in most other directions. By using this feature, only the channels on the non-zero components can be fed back, which greatly reduces the overhead compared to directly feeding back all channels. Therefore, the beams in the angle domain can be compressed in the spatial domain. In the Rel15 version, the TypeII codebook compresses the eigenvector in the spatial domain through L DFT beams to obtain W 2 , and in the Rel16 version, W 2 The DFT orthogonal matrix is ​​compressed in the frequency domain to obtain Thus, the feedback overhead can be further reduced, and a higher number of transmission layers can be supported. In the calculation process of the precoding matrix of Type II of Rel15 and enhanced Type II of Rel16 in the prior art, when W is calculated 1 After that, the equivalent channel correlation matrix of all subbands is calculated first, and then the eigenvalue decomposition of all subbands is performed to obtain W2 , and finally to W 2 Quantification is performed to obtain the final reported W 2 This process needs to be performed for all sub-bands. For example, in Rel15 Type II, the number of sub-bands is 19, while in Rel16e Type II, the number of sub-bands is 37. Therefore, the complexity of this step is very high.

[0004] To this end, in view of the shortcomings of the prior art solutions, the present invention mainly targets Type II W 1 The selection process provides a precoding matrix selection method suitable for Type II, which solves the problem of Type II calculation process being complicated and due to W 1 Inaccurate selection may lead to performance degradation. Summary of the invention

[0005] The present invention aims to overcome at least one defect (deficiency) of the above-mentioned prior art and provide a precoding matrix selection method applicable to Type II, which is used to solve the problem that the calculation process of Type II is complicated and due to W 1 Inaccurate selection leads to problems such as performance degradation.

[0006] The technical solution adopted by the present invention is a method for selecting a precoding matrix applicable to Type II, the method comprising the following steps:

[0007] S1: Calculate N 1 O 1 N 2 O 2 The power of the beam P beam (i),0≤i <N 1 O 1 N 2 O 2 ;

[0008] S2: Compare the CSI-RS signal noise ratio with the preset threshold value to determine the current CSI-RS channel quality;

[0009] S3: Select a suitable beam scheme based on the judgment result of step S2 to calculate the optimal candidate beam W 1 When the CSI-RS signal noise ratio is less than the preset threshold value, it means that the CSI-RS channel quality is poor and there is a large error in the CSI-RS channel estimation. The scheme of selecting the main beam is adopted; when the CSI-RS signal noise ratio is greater than the preset threshold value, it means that the CSI-RS channel quality is good and the CSI-RS channel estimation is accurate. The scheme of selecting multiple candidate beam combinations is adopted.

[0010] In this application, for Type II W 1In the selection process, different implementation schemes are selected according to the CSI-RS signal quality. When the signal-to-noise ratio is low, there are certain errors in channel estimation, and the power calculation of each beam is inaccurate, but the calculation of the strongest beam will be more accurate. Therefore, when the signal-to-noise ratio is low, the main beam with the largest power can be selected, and the beam orthogonal to it can be found based on the main beam. It is ensured that the strongest beam must be in the candidate beam combination, which can ensure the robustness of the system and reduce the complexity of the implementation. When the signal-to-noise ratio is high, if the W 1 Selecting only one candidate set in the selection phase will lead to system performance degradation, especially in non-direct path channel environments where beams come from many angles. Selecting only one candidate beam is not enough. Therefore, 1 In the selection phase, multiple candidate beam combinations are selected. In order to avoid multiple W 1 Bring it into subband W 2 During the selection process, the optimal beam is selected in the dimension of the wideband channel correlation matrix, and it is not substituted into the subband module for selection, which can improve the performance of the system in a scenario with limited increase in complexity. Therefore, the present invention can greatly reduce the implementation complexity when the performance loss is small by executing different beam selection schemes according to different channel qualities, and effectively improve the system performance.

[0011] Preferably, in step S3, the scheme of selecting the main beam includes the following steps:

[0012] S4: From N 1 O 1 N 2 O 2 Select the beam index with the largest power among the beams and calculate the corresponding beam parameter combination n 1,2,1,2 ;

[0013] S5: Select N that is orthogonal to the main beam 1 N 2 -1 beam and the main beam form a candidate beam set;

[0014] S6: From N 1 N 2 Select the L beams with the largest power from the beam set as the optimal candidate beam W 1 .

[0015] In the present application, when the signal-to-noise ratio is low, by selecting the main beam with the largest power and searching for a beam orthogonal to the main beam as a reference, it is ensured that the strongest beam is in the candidate beam combination, thereby ensuring the robustness of the system and reducing the complexity of implementation.

[0016] Preferably, in step S4, the beam index with the largest power and the corresponding beam parameter combination n 1,2,1,2The calculation formula is:

[0017] maxIdx=argmax{i|P beam (i)},0≤i< 1 O 1 N 2 O 2 ;

[0018]

[0019] q 1 =-n 1 O 1,1 ∈{0,1,…,O 1 -1{;

[0020] m=maxIdx-· 2 O 2 ,∈{0,1,…,N 2 O 2 -1};

[0021]

[0022] q 2 =-n 2 O 2,2 ∈{0,1,…,O 2 -1};

[0023] Where maxIdx represents the beam index with the largest power, N 1 Indicates the number of antennas in the horizontal direction, N 2 Indicates the number of antennas in the vertical direction, O 1 Indicates the horizontal beam oversampling multiple, O 2 Indicates the vertical beam oversampling multiple, n 1 represents the orthogonal beam index in the horizontal direction, q 1 Indicates the horizontal oversampling beam index, n 2 represents the orthogonal beam index in the vertical direction, q 2 represents the oversampling beam index in the vertical direction, l represents the beam index in the horizontal direction, and m represents the beam index in the vertical direction.

[0024] By calculating the beam parameter combination corresponding to the beam index with the highest power, the foundation can be laid for the subsequent calculation of the optimal main beam, thereby improving the signal quality and transmission reliability.

[0025] Preferably, in step S5, the set is:

[0026] idx=mod((n 1 + 1 ) 1+ 1 +(n 2 + 2 ) 2 + 2,1 O 1 N 2 O 2 );

[0027] k 1 =0,1…,N 1 -1,k 2 =0,1…,N 2 -1;

[0028] Among them, N 1 Indicates the number of antennas in the horizontal direction; N 2 Indicates the number of antennas in the vertical direction.

[0029] Preferably, in step S3, the scheme of selecting multiple candidate beam combinations comprises the following steps:

[0030] S7: For each group (q 1,2 )Combined calculation of all (n 1,2 ) combine the L beam combinations with the largest combined powers and record the corresponding index values ​​of the L beam combinations and the maximum beam power sum of the L beam combinations;

[0031] S8: From N 1 O 1 N 2 O 2 Select N values ​​with the largest power sum from the combinations;

[0032] S9: For each group Compute the equivalent broadband correlation matrix

[0033] S10: Yes Find the maximum v max Eigenvalue λ i ;

[0034] S11: Calculate each candidate set W 1,c The corresponding broadband metric value;

[0035] S12: Select W corresponding to the largest broadband metric value 1 .

[0036] In this application, when the signal-to-noise ratio is high, if W 1Selecting only one candidate set during the selection phase will lead to a decrease in system performance, especially in non-direct path channel environments. At this time, the beams come from many angles, and it is not enough to select only one candidate beam. Therefore, multiple candidate sets need to be selected during beam selection, and the optimal beam is selected in the broadband domain without being substituted into the subband module for selection. This can improve system performance in scenarios with limited increase in complexity.

[0037] Preferably, the step S7 includes the following formula:

[0038]

[0039] in, It means that when q=(q 1,2 ), the L largest beam indexes, and the actual beam index is Total N 1 O 1 N 2 O 2 Combinations, 0≤q 1 <O 1 ,0≤q 2<2 ;

[0040]

[0041] Among them, P bc (n 1,2,1,2 ) represents the power sum of L beam combinations, P beam (idx l ) represents the beam power of each real beam index.

[0042] In the present application, by selecting L optimal beam combinations through the above formula and recording their corresponding beam index values ​​and maximum beam power sum, the signal transmission path can be optimized in multiple different environmental conditions and target areas to avoid the influence of signal attenuation, interference or shielding; and in a network with limited resources, the L beam combinations with the largest power can be accurately selected, and spectrum resources can be allocated more efficiently. The system can maximize the utilization of its resources and provide stable and efficient services to more users; in addition, it can also make decisions quickly during real-time scheduling, reduce calculation delays, and improve system response speed and real-time performance.

[0043] Preferably, in step S9, the equivalent broadband correlation matrix The formula is:

[0044]

[0045] Where R is the wideband channel correlation matrix of the channel, H is the channel estimation value after whitening each sample point, W1,c represents the candidate set, represents the conjugate transpose of the precoding matrix.

[0046] By calculating the equivalent wideband correlation matrix, the optimal beam can be selected in the wideband domain without substituting it into the subband module for selection. This can improve the performance of the system in scenarios with limited increase in complexity.

[0047] Preferably, in step S10, only the eigenvalue needs to be solved without solving the eigenvector, and the formula is:

[0048]

[0049] Let λ c,i Arrange them in order from largest to smallest, i.e. Then select the largest V according to this sorting max Eigenvalue λ i .

[0050] When selecting beams in the broadband domain, it is only necessary to solve the maximum number of layers V max The eigenvalue of is sufficient without solving the eigenvector, which effectively reduces the computational complexity.

[0051] Preferably, in step S11, the calculation formula of the broadband metric value is:

[0052]

[0053] Among them, ρ c Indicates the bandwidth metric.

[0054] Preferably, the step S12 includes:

[0055]

[0056] in, Represents the index of the candidate precoding matrix with the largest wideband metric value.

[0057] By calculating the broadband metric value, the optimal beam selection can be completed in the broadband domain, which helps to accurately control the beam, reduce interference, and improve system performance when the complexity increases.

[0058] Compared with the prior art, the present invention has the following beneficial effects:

[0059] This application is mainly for Type II W 1In the selection process, different implementation schemes are selected according to the CSI-RS signal quality. When the signal-to-noise ratio is low, there are certain errors in channel estimation, and the power calculation of each beam is inaccurate, but the calculation of the strongest beam will be more accurate. Therefore, when the signal-to-noise ratio is low, the main beam with the largest power can be selected, and the beam orthogonal to it can be found based on the main beam. It is ensured that the strongest beam must be in the candidate beam combination, which can ensure the robustness of the system and reduce the complexity of the implementation. When the signal-to-noise ratio is high, if the W 1 Selecting only one candidate set in the selection phase will lead to system performance degradation, especially in non-direct path channel environments where beams come from many angles. Selecting only one candidate beam is not enough. Therefore, 1 In the selection phase, multiple candidate beam combinations are selected. In order to avoid multiple W 1 Bring it into subband W 2 During the selection process, the optimal beam is selected in the dimension of the wideband channel correlation matrix, and it is not substituted into the subband module for selection, which can improve the performance of the system in a scenario with limited increase in complexity. Therefore, the present invention can greatly reduce the implementation complexity when the performance loss is small by executing different beam selection schemes according to different channel qualities, and effectively improve the system performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 This is a flow chart of the precoding matrix selection method provided by the present invention.

[0061] Figure 2 This is a structural diagram of the NR antenna provided by the present invention.

[0062] Figure 3 The W provided by the present invention 1 Schematic diagram of correlation for the same polarization direction.

[0063] Figure 4 The W provided by the present invention 1 Schematic diagram of the beam.

[0064] Figure 5 This is the codebook evolution process of Type II.

[0065] Figure 6 The block diagram of Type II codebook selection for the existing scheme.

[0066] Figure 7 A flow chart of the solution for selecting the main beam provided by the present invention.

[0067] Figure 8 A flow chart of a solution for selecting multiple candidate beam combinations provided by the present invention.

[0068] Fig. 9This is a schematic diagram of the high signal-to-noise ratio simulation results provided by the present invention.

[0069] Fig.10 This is a schematic diagram of the low signal-to-noise ratio simulation results provided by the present invention. DETAILED DESCRIPTION

[0070] The drawings of the present invention are only for illustrative purposes and should not be construed as limiting the present invention. In order to better illustrate the following embodiments, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of the actual product; it is understandable to those skilled in the art that some well-known structures and their descriptions in the drawings may be omitted.

[0071] Example 1

[0072] like Figure 1 As shown, this embodiment provides a method for selecting a precoding matrix applicable to Type II, and the method includes the following steps:

[0073] Step S1: Calculate N 1 O 1 N 2 O 2 The power of the beam P beam (i),0≤i <N 1 O 1 N 2 O 2 ;

[0074] Step S2: Compare the CSI-RS signal noise ratio with a preset threshold value to determine the channel quality of the current CSI-RS;

[0075] Step S3: Select a suitable beam solution according to the judgment result of step S2 to calculate the optimal candidate beam W 1 When the CSI-RS signal noise ratio is less than the preset threshold value, it means that the CSI-RS channel quality is poor and there is a large error in the CSI-RS channel estimation. The scheme of selecting the main beam is adopted; when the CSI-RS signal noise ratio is greater than the preset threshold value, it means that the CSI-RS channel quality is good and the CSI-RS channel estimation is accurate. The scheme of selecting multiple candidate beam combinations is adopted.

[0076] In the NR system, when the number of base station antenna ports is greater than 2, dual-polarized antennas are used, where the antenna configuration is defined as (N 1,2 ), N 1 Indicates the number of antennas in the horizontal direction, N 2 Indicates the number of antennas in the vertical direction. Figure 2 As shown, Figure 2The NR antenna structure diagram provided by the present invention, in which all red antennas represent +45° polarization direction antennas, and all blue antennas represent -45° polarization direction antennas. 1 =4,N 2 =2, then P = 2 × N 1 ×N 2 =16.

[0077] Secondly, all codebook designs adopt a parameterized codebook structure, using a two-level codebook W = W 1 · 2 , where W 1 Describes the long-term broadband characteristics of the channel, W 2 Describes the short-term subband characteristics of the channel. 1 It is composed of several orthogonal beams, each of which can be represented by b, consisting of horizontal beams and vertical beam Kronecker product. Figure 3 and Figure 4 As shown, Figure 3 W 1 Schematic diagram of correlation in the same polarization direction. Figure 4 W 1 Schematic diagram of the beam. Among them, W 1 The expressions of a and b are as follows:

[0078]

[0079] Figure 4 The 8 red dots are orthogonal basic beams, the blue dots represent oversampled beams, and the horizontal beam index is l=O. 1 · 1 + 1 ≤n 1<1 -1,0≤q 1<1 -1, the vertical beam index is m=0 2 · 2 +q 2 ,0≤n 2<2 -1,0≤q 2<2 -1,O 1 Indicates the horizontal beam oversampling multiple, O 2 Indicates the beam oversampling multiple in the vertical direction.

[0080] NR supports two types of codebooks for CSI feedback, namely Type I codebooks with normal precision and Type II codebooks with high precision. For Type II, the basic principle is to transform the spatial channel into the angle domain, and the combined coefficients represent the amplitude and phase of each beam component. On the one hand, the channel itself is sparse in space, with energy in only a few directions and zero energy in most other directions. By using this feature, only the channels on the non-zero components can be fed back, which greatly reduces the overhead compared to directly feeding back all channels. Therefore, the beams in the angle domain can be compressed in the spatial domain, W 1 represents a non-zero orthogonal beam in the spatial domain, with dimensions P×2L, W 2 It represents the coefficients (including amplitude and phase) of each beam corresponding to each layer, with a dimension of 2×v. W represents the precoding matrix, with a dimension of P×v.

[0081] like Figure 5 As shown, Figure 5 This is the evolution process of the Type II codebook. In the Rel15 version, the Type II codebook compresses the feature vector in the spatial domain through L DFT beams to obtain W 2 , in Rel16, W 2 The DFT orthogonal matrix is ​​compressed in the frequency domain to obtain Thereby, the feedback overhead can be further reduced, and a higher number of transmission layers can be supported. 1 This solution is applicable to both Rel15 Type II and Rel16 enhanced Type II.

[0082] The calculation process of the precoding matrix of Type II and eType II in the prior art solution is as follows, where W 1 The generation of Figure 6 The selection process of W1 in the gray box is shown. Figure 6 The block diagram of Type II codebook selection for the existing solution is shown below:

[0083] Step 1: Calculate N 1 O 1 N 2 O 2 The power of the beam P beam (i),0≤i< 1 O 1 N 2 O 2 ;

[0084] Step 2: For each group (q 1,2 )Combined calculation of all (n 1,2 ) Combine the L beams with the largest combined power and record the corresponding L beam index values ​​and the maximum beam power sum of the L beam combinations;

[0085]

[0086] In the formula, It means that when q=(q 1,2 ), the L largest beam indexes, and the actual beam index is Total N 1 O 1 N 2 O 2 Combinations, 0≤q 1 <O 1 ,0≤q 2<2 .

[0087]

[0088] Step 3: From N 1 O 1 N 2 O 2 The combination of these two types selects the one with the largest power and the largest value as the candidate beam W. 1 ;

[0089]

[0090] Step 4: Calculate W 1 After that, the equivalent channel correlation matrix of all subbands is calculated first, and then the eigenvalue decomposition of all subbands is performed to obtain W 2 , and then to W 2 Quantify and obtain the final reported W 2 .

[0091] Specifically, in the prior art, multiple candidate sets W are selected. 1 , then calculate the equivalent correlation matrix of all subbands, perform EVD decomposition on it, obtain the subband set PMI, calculate the MI value and of all sample points based on PMI, and then select the largest W 1 As the final reported PMI, this solution has better performance, but it is highly complex and requires calculation of W corresponding to all subbands. 2 , which means that EVD decomposition is performed on all subbands, which is equivalent to traversing all candidate beams in step 4 of the existing implementation scheme, so the complexity is a multiple of the candidate beam value of the existing implementation scheme. In addition, it should be noted that step 4 needs to be performed on all subbands. For example, in Rel15 TypeII, the number of subbands is 19, and in Rel16 eTypeII, the number of subbands is 37. It can be seen that the complexity of step 4 of the prior art is very high.

[0092] To this end, in view of the shortcomings of the existing technical solutions, in order to reduce the implementation complexity of step 4 and solve the problem of W 1 Inaccurate selection leads to performance degradation. This embodiment proposes the following Figure 1 The precoding matrix selection method applicable to Type II is shown to solve the above problem. This method is mainly for Type II W 1 In the selection process, different implementation schemes are selected according to the CSI-RS signal quality. When the signal-to-noise ratio is low, there are certain errors in channel estimation, and the power calculation of each beam is inaccurate, but the calculation of the strongest beam will be more accurate. Therefore, when the signal-to-noise ratio is low, the main beam with the largest power can be selected, and the beam orthogonal to it can be found based on the main beam. It is ensured that the strongest beam must be in the candidate beam combination, which can ensure the robustness of the system and reduce the complexity of the implementation. When the signal-to-noise ratio is high, if the W 1 Selecting only one candidate set in the selection phase will lead to system performance degradation, especially in non-direct path channel environments where beams come from many angles. Selecting only one candidate beam is not enough. Therefore, 1 In the selection phase, multiple candidate beam combinations are selected. In order to avoid multiple W 1 Bring it into subband W 2 During the selection process, the optimal beam is selected in the dimension of the wideband channel correlation matrix, and it is not substituted into the subband module for selection, which can improve the performance of the system in a scenario with limited increase in complexity. Therefore, the present invention can greatly reduce the implementation complexity when the performance loss is small by executing different beam selection schemes according to different channel qualities, and effectively improve the system performance.

[0093] Preferably, in step S2, for the CSI-RS signal noise ratio, its formula can be expressed as:

[0094]

[0095] Among them, P signal is the power of the received CSI-RS signal; P noise is the received noise power.

[0096] After the CSI-RS signal noise ratio is obtained by using the above formula, it is compared with a preset threshold value to evaluate the quality of the channel, wherein the numerical value of the threshold value can be set by those skilled in the art according to actual needs.

[0097] Preferably, if Figure 7 As shown, in step S3, when the CSI-RS signal noise ratio is less than the preset threshold value, the scheme of selecting the main beam includes the following steps:

[0098] Step S4: From N 1 O 1 N 2 O 2 Select the beam index with the largest power among the beams and calculate the corresponding beam parameter combination n 1,2,1,2 ;

[0099] Preferably, in step S4, the beam index with the largest power and the corresponding beam parameter combination n 1,2,1,2 The calculation formula is:

[0100] maxIdx=argmax{i|P beam (i)},0≤i< 1 O 1 N 2 O 2 ;

[0101]

[0102] q 1 =-n 1 O 1,1 ∈{0,1,…,O 1 -1};

[0103] m=maxIdx-· 2 O 2 ,∈{0,1,…,N 2 O 2 -1};

[0104]

[0105] q 2 =-n 2 O 2,2 ∈{0,1,…,O 2 -1};

[0106] Where maxIdx represents the beam index with the largest power, N 1 Indicates the number of antennas in the horizontal direction, N 2 Indicates the number of antennas in the vertical direction, O 1 Indicates the horizontal beam oversampling multiple, O 2 Indicates the vertical beam oversampling multiple, n 1 represents the orthogonal beam index in the horizontal direction, q 1 Indicates the horizontal oversampling beam index, n 2 represents the orthogonal beam index in the vertical direction, q 2 represents the oversampling beam index in the vertical direction, l represents the beam index in the horizontal direction, and m represents the beam index in the vertical direction.

[0107] By calculating the beam parameter combination corresponding to the beam index with the highest power, the foundation can be laid for the subsequent calculation of the optimal main beam, thereby improving the signal quality and transmission reliability.

[0108] Step S5: Select N beams orthogonal to the main beam 1 N 2 -1 beam and the main beam form a candidate beam set;

[0109] Preferably, in step S5, the set is:

[0110] idx=mod((n 1 + 1 ) 1 + 1 +(n 2 + 2 ) 2 + 2,1 O 1 N 2 O 2 );

[0111] k 1 =0,1…,N 1 -1,k 2 =0,1…,N 2 -1;

[0112] Among them, N 1 Indicates the number of antennas in the horizontal direction; N 2 Indicates the number of antennas in the vertical direction.

[0113] Step S6: From N 1 N 2 Select the L beams with the largest power from the beam set as candidate beams W 1 .

[0114]

[0115] Therefore, in this embodiment, when the signal-to-noise ratio is low, by selecting the main beam with the largest power and searching for a beam orthogonal to the main beam as a reference, it is ensured that the strongest beam is in the candidate beam combination, thereby ensuring the robustness of the system and reducing the complexity of implementation.

[0116] Preferably, if Figure 8 As shown, in step S3, when the CSI-RS signal noise ratio is greater than a preset threshold value, the scheme of selecting multiple candidate beam combinations includes the following steps:

[0117] Step S7: For each group 9q 1,2 ) Combine and calculate all 9n 1,2) combine the L beam combinations with the largest combined powers and record the corresponding index values ​​of the L beam combinations and the maximum beam power sum of the L beam combinations;

[0118] Preferably, the step S7 includes the following formula:

[0119]

[0120] in, It means that when q=(q 1,2 ), the L largest beam indexes, and the actual beam index is Total N 1 O 1 N 2 O 2 Combinations, 0≤q 1 <O 1 ,0≤q 2<2 ;

[0121]

[0122] Among them, P bc (n 1,2,1,2 ) represents the power sum of L beam combinations, P beam (idx l ) represents the beam power of each real beam index.

[0123] In the present application, by selecting L optimal beam combinations through the above formula and recording their corresponding beam index values ​​and maximum beam power sum, the signal transmission path can be optimized in multiple different environmental conditions and target areas to avoid the influence of signal attenuation, interference or shielding; and in a network with limited resources, the L beam combinations with the largest power can be accurately selected, and spectrum resources can be allocated more efficiently. The system can maximize the utilization of its resources and provide stable and efficient services to more users; in addition, it can also make decisions quickly during real-time scheduling, reduce calculation delays, and improve system response speed and real-time performance.

[0124] Step S8: From N 1 O 1 N 2 O 2 Select N values ​​with the largest power sum from the combinations;

[0125] Preferably, the formula is:

[0126]

[0127] In this embodiment, the typical value N=2 is selected to obtain the maximum value of the power sum, as shown below:

[0128]

[0129] Step S9: For each group Compute the equivalent broadband correlation matrix

[0130] Preferably, in step S9, the equivalent broadband correlation matrix The formula is:

[0131]

[0132] Where R is the wideband channel correlation matrix of the channel, H is the channel estimation value after whitening each sample point, 1,c represents the c-th candidate precoding matrix, represents the conjugate transpose of the precoding matrix.

[0133] By calculating the equivalent wideband correlation matrix, the optimal beam can be selected in the wideband domain without substituting it into the subband module for selection. This can improve the performance of the system in scenarios with limited increase in complexity.

[0134] Step S10: Find the maximum V max Eigenvalue λ i ;

[0135] Preferably, in step S10, only the eigenvalue needs to be solved without solving the eigenvector, and the formula is:

[0136]

[0137] Let λ c,i Arrange in order from large to small, that is, λ c,0 ≥λ c,1 ≥…≥λ c,Vmax-1 , and then select the largest V according to this sorting max Eigenvalue λ i .

[0138] When selecting beams in the broadband domain, it is only necessary to solve the maximum number of layers V max The eigenvalue of is sufficient without solving the eigenvector, which effectively reduces the computational complexity.

[0139] Step S11: Calculate each candidate set W 1,c The corresponding broadband metric value;

[0140] Preferably, in step S11, the calculation formula of the broadband metric value is:

[0141]

[0142] Among them, ρ c Indicates the bandwidth metric.

[0143] Step S12: Select W corresponding to the maximum broadband metric value 1 .

[0144] Preferably, the step S12 includes:

[0145]

[0146] in, Indicates the index of the candidate precoding matrix with the largest wideband metric value.

[0147] By calculating the broadband metric, the optimal beam selection can be completed in the broadband domain, which helps to accurately control the beam, reduce interference, and improve system performance when the complexity increases. When selecting beams in the broadband domain, it is assumed that the maximum number of layers that the system can support is V max , then the calculation can be Maximum V max feature values, select The largest beam is used as the candidate beam in the broadband domain.

[0148] Therefore, in this embodiment, when the signal-to-noise ratio is high, if W 1 Selecting only one candidate set during the selection phase will lead to a decrease in system performance, especially in non-direct path channel environments. At this time, the beams come from many angles, and it is not enough to select only one candidate beam. Therefore, multiple candidate sets need to be selected during beam selection, and the optimal beam is selected in the broadband domain without being substituted into the subband module for selection. This can improve system performance in scenarios with limited increase in complexity.

[0149] In addition, in order to verify the performance of the present solution, this embodiment also provides simulation comparisons of the performance gain of the present solution relative to the existing solution for two scenarios, low signal-to-noise ratio and high signal-to-noise ratio. For high signal-to-noise ratio, the simulation parameters are set as shown in Table 1. The simulation results of the present solution and the existing solution are shown in Table 1. Fig. 9 As shown; for low signal-to-noise ratio, the simulation parameters set are shown in Table 2, and the simulation results of the scheme of the present invention and the existing implementation scheme are shown in Fig.10 shown.

[0150] Simulation parameters value Number of PRBs 48 Number of base station antennas 8 Number of UE receiving antennas 4 CSI-RS period 5ms MCS Rank2,MCS20 Channel Model TDL-A Movement Speed 3km / h

[0151] Table 1

[0152] Simulation parameters value Number of PRBs 48 Number of base station antennas 8 Number of UE receiving antennas 4 CSI-RS period 5ms MCS Rank1,MCS8 Channel Model TDL-A Movement Speed 3km / h

[0153] Table 2

[0154] from Fig. 9 and Fig.10 From the simulation results, it can be seen that under low signal-to-noise ratio, the performance gain of the scheme using the main beam is about 0.3dB compared with the existing technical scheme, while in a multipath environment with high signal-to-noise ratio and rich scattering paths, the performance gain of selecting multiple candidate beams is about 0.6dB compared with the existing technical scheme.

[0155] It can be seen that the optimal W is obtained by selecting the main beam or performing eigenvalue decomposition on the equivalent correlation matrix in the broadband domain disclosed in this embodiment. 1 The technical solution is obviously superior to the existing technical solution, and can greatly reduce the implementation complexity with less performance loss. It only needs to calculate the optimal candidate set in the broadband domain without traversing all candidate beams, which effectively reduces the calculation complexity.

[0156] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the technical solution of the present invention, and are not intended to limit the specific implementation methods of the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the claims of the present invention shall be included in the protection scope of the claims of the present invention.

Claims

1. A method for selecting a precoding matrix applicable to Type II, characterized in that: The method comprises the following steps: S1: Calculate the power P of the N1O1N2O2 beams beam (i),0≤i <N1O1N2O2; S2: Compare the CSI-RS signal noise ratio with the preset threshold value to determine the current CSI-RS channel quality; S3: According to the judgment result of step S2, a suitable beam scheme is selected to calculate the optimal candidate beam W1. When the signal-to-noise ratio of the CSI-RS signal is less than the preset threshold value, it means that the quality of the CSI-RS channel is poor and there is a large error in the CSI-RS channel estimation. The scheme of selecting the main beam is adopted; when the signal-to-noise ratio of the CSI-RS signal is greater than the preset threshold value, it means that the quality of the CSI-RS channel is good and the CSI-RS channel estimation is accurate. The scheme of selecting multiple candidate beam combinations is adopted.

2. A method for selecting a precoding matrix suitable for Type II according to claim 1, characterized in that: In step S3, the scheme of selecting the main beam includes the following steps: S4: Select the beam index with the largest power from the N1O1N2O2 beams and calculate the corresponding beam parameter combination n1, n2, q1, q2; S5: Select N1N2-1 beams orthogonal to the main beam and the main beam to form a candidate beam set; S6: Select L beams with the largest power from the N1N2 beam sets as the optimal candidate beams W1.

3. The method for selecting a precoding matrix applicable to Type II according to claim 2, characterized in that: In step S4, the calculation formula of the beam index with the maximum power and the corresponding beam parameter combination n1, n2, q1, q2 is: maxIdx=argmax{i|P beam (i)},0≤i<N1O1N2O2; q1=l-n1O1, q1∈{0,1,…,O1-1}; m=maxIdx-l·N2O2,m∈{0,1,…,N2O2-1}; q2=m-n2O2,q2∈{0,1,…,O2-1}; Among them, maxIdx represents the beam index with the largest power, N1 represents the number of antennas in the horizontal direction, N2 represents the number of antennas in the vertical direction, O1 represents the beam oversampling multiple in the horizontal direction, O2 represents the beam oversampling multiple in the vertical direction, n1 represents the orthogonal beam index in the horizontal direction, q1 represents the oversampled beam index in the horizontal direction, n2 represents the orthogonal beam index in the vertical direction, q2 represents the oversampled beam index in the vertical direction, l represents the beam index in the horizontal direction, and m represents the beam index in the vertical direction.

4. The method for selecting a precoding matrix applicable to Type II according to claim 3, characterized in that: In step S5, the set is: idx=mod((n1+k1)·O1+q1+(n2+k2)·O2+q2,N1O1N2O2); k1=0,1…,N1-1,k2=0,1…,N2-1; Wherein, N1 represents the number of antennas in the horizontal direction; N2 represents the number of antennas in the vertical direction.

5. A method for selecting a precoding matrix applicable to Type II according to any one of claims 1 to 4, characterized in that: In step S3, the scheme of selecting multiple candidate beam combinations includes the following steps: S7: for each group (q1, q2) of combinations, calculate the L beam combinations with the largest combined power of all (n1, n2) combinations and record the corresponding index values ​​of the L beam combinations and the maximum beam power sum of the L beam combinations; S8: Select N values ​​with the largest power sum from the combinations of N1O1N2O2; S9: For each group Compute the equivalent broadband correlation matrix S10: Yes Find the maximum V max Eigenvalue λ i ; S11: Calculate each candidate set W 1,c The corresponding broadband metric value; S12: Select W1 corresponding to the maximum broadband metric value.

6. A method for selecting a precoding matrix applicable to Type II according to claim 5, characterized in that: The step S7 includes the following formula: in, It means that when q = (q1, q2), there are L largest beam indices. The real beam index is There are N1O1N2O2 combinations, 0≤q1 <O1,0≤q2<O2; Among them, P bc (n1,n2,q1,q2) represents the power sum of L beam combinations, P beam (idx l ) represents the beam power of each real beam index.

7. A method for selecting a precoding matrix applicable to Type II according to claim 6, characterized in that: In step S9, the effective broadband correlation matrix The formula is: Where R is the wideband channel correlation matrix of the channel, H is the channel estimation value after whitening for each sample point, W 1,c represents the c-th candidate precoding matrix, represents the conjugate transpose of the precoding matrix.

8. A method for selecting a precoding matrix applicable to Type II according to claim 7, characterized in that: In step S10, only the eigenvalue needs to be solved, and the eigenvector does not need to be solved. The formula is: Let λ c,i Arrange them in order from largest to smallest, i.e. Then select the largest V according to this sorting max Eigenvalue λ i .

9. A method for selecting a precoding matrix applicable to Type II according to claim 8, characterized in that: In step S11, the calculation formula of the broadband metric value is: Among them, ρ c Indicates the bandwidth metric.

10. A method for selecting a precoding matrix applicable to Type II according to claim 9, characterized in that: The step S12 includes: in, Indicates the index of the candidate precoding matrix with the largest wideband metric value.

Images