CN106788631A - A kind of extensive MIMO reciprocities calibration method based on local alignment - Google Patents

Abstract

The present invention proposes the extensive MIMO reciprocities calibration method based on local alignment under a kind of TDD mode, this method is applied to centralized antenna scene and spaced antenna scene, it is different antenna sets by by the adjacent antenna partition in base station, different antenna sets are carried out to be calibrated between local alignment and group in group, final to realize overall calibration, the advantage of the algorithm is：Without the feedback from user terminal；The number of antennas that LS calibrations are related to is limited in certain scope.In the case of calibration accuracy loss very little, it is ensured that calibration complexity is substantially reduced.

Description

A kind of extensive MIMO reciprocities calibration method based on local alignment

Technical field

The invention belongs to communication technical field, more particularly to a kind of extensive MIMO reciprocities calibration based on local alignment Method.

Background technology

Due to the characteristic, the application day of extensive MIMO technology such as radio frequency link module low cost, anti-fading, capacity be big Beneficial extensive, it has become Future cellular networks, the alternative technique of 5G communications.In practical communication, extensive mimo system is past Toward needing to obtain descending channel information (CSI), to carry out precoding, systemic-function is realized.

In time division duplex (TDD) pattern, send and receive channel and be carried in identical carrier frequency, use guard time It is spaced to realize the separation of up-downgoing channel.In overall TDD system, the channel between base station and user uses identical carrier wave Frequency, therefore the uplink downlink characteristic of channel is identical, can by uplink channel estimation to channel gain be directly used in descending letter Road precoding.

Although physical channel has reciprocity, in practical communication, channel transmitting and reception process be required for by Radio frequency link module, every antenna is received and the radio frequency link of transmission is respectively completed by different circuits, is launched and is received The characteristic of circuit is difficult to accomplish consistent.Further, since the characteristic such as transmitting and environment temperature and humidity residing for receiving circuit is different, Two sets of characteristics of circuit can not accomplish identical so that channel reciprocity is damaged, and needs to carry out channel in practical communication Reciprocity is calibrated.

Known reciprocity calibration algorithm employs global calibration program, they by all of antenna be considered as one it is whole Body.As number of antennas is increasing, the treatment expense of global calibration program will be often very big.

But when using more simple collimation technique, although complexity has very big decline, but calibration accuracy is damaged Lose very big.

The content of the invention

In order to overcome the shortcoming of above-mentioned prior art, it is an object of the invention to provide a kind of big rule based on local alignment Mould MIMO reciprocity calibration methods, its loss of significance is small, and complexity is low, it is adaptable to the extensive mimo system of time division duplex, it Adaptable scene is：(1) centralized antenna scene；(2) spaced antenna scene.

To achieve these goals, the technical solution adopted by the present invention is：

A kind of extensive MIMO reciprocities calibration method based on local alignment, it is adaptable to the extensive MIMO of time division duplex System, comprises the following steps：

All antenna for base station are divided into different groups by step A, and each group of base stations antenna is calibrated using LS algorithms, are obtained Obtain the local alignment coefficient of each group antenna for base station；

Step B, to different groups, selects some foundation station antennas as reference antenna, is calibrated between obtaining different groups of group Coefficient；

Step C, using calibration factor between the local alignment coefficient and group obtained in step A and B, is calculated global calibration Coefficient c=(c₁,c₂,...,c_M)^T, c_i=r_i/t_i(i=1,2 ..., M), M is the sum of antenna for base station, t_iIt is antenna for base station i Radiofrequency emitting module Equivalent Base-Band gain, r_iIt is the Equivalent Base-Band gain of the Receiver Module of antenna for base station i；

Step D, using global calibration factor, the transmitting and receiving module to antenna for base station is adjusted, and realizes relative calibration Process.

In step A, antenna for base station is fifty-fifty divided into m groups as far as possible, while setting threshold value d₀, it is ensured that every group of inside day The distance between line is less than d₀。

It is as follows the step of calibrated using LS algorithms in step A：

Step 1, in antenna sets, mutually sends pilot tone between any two foundation station antenna, measure any two base station Channel gain between antenna

Step 2, using the channel gain between any two foundation station antennaStructural matrix W：

Step 3, Eigenvalues Decomposition is carried out to matrix W, obtains the characteristic vector corresponding to its minimal eigenvalue, as locally Calibration factorWherein i and j is the sequence number of antenna for base station, and m is the sequence number of antenna sets, M_mIt is m groups day The number of line.

In step B, for antenna sets m₁With antenna sets m₂When being calibrated between group, it is assumed that choose n roots antenna altogether as referring to day Line, then from antenna sets m₁Middle selected distance antenna sets m₂Nearest n/2 root antennas, from antenna sets m₂Middle selected distance antenna sets m₁ Nearest n/2 root antennas, constitute a new antenna sets

In step B, when being calibrated using adjacent antenna group recursion, some base station are only chosen from the antenna sets of adjacent antenna Antenna constitutes a new new antenna setsNew antenna sets are carried out with LS calibrations described in claim 3, is obtained not on the same day Calibration factor between the group of line group

In step B, during using single calibration with reference to group, antenna sets m is taken_n/2As reference antenna group, selected from other antenna sets Take some foundation station antennas and antenna sets m_n/2In some base station antenna sets into the new antenna sets of antennaTo new day Line group carries out LS calibrations described in claim 3, obtains calibration factor between the group of different antennae group

Compared with prior art, the beneficial effects of the invention are as follows：

(1) calibration process can be completed using the channel response between antenna for base station, without extra hardware circuit, without Feedback from user terminal.

(2) by setting the size in local alignment region, on the premise of ensureing that calibration accuracy is not adversely effected, calibration Complexity is substantially reduced.

Brief description of the drawings

Fig. 1 is the extensive mimo system structural representation of time division duplex according to one embodiment of the invention.

Fig. 2 is the linear array list based on local alignment according to one embodiment of the invention with reference to group calibration method structure chart.

Fig. 3 is the linear array adjacent sets recursion calibration method structure based on local alignment according to one embodiment of the invention.

Fig. 4 is calibration program performance evaluation between two kinds of groups of linear array that local alignment is based on according to one embodiment of the invention Analogous diagram.

Fig. 5 is based on local alignment face battle array adjacent sets recursion calibration structure figure according to one embodiment of the invention.

Fig. 6 is the global calibration program of face battle array and the local alignment side that local alignment is based on according to one embodiment of the invention Case performance simulation figure.

Specific embodiment

Embodiments of the invention are described below in detail, the example of embodiment is shown in the drawings, wherein identical from start to finish Or similar label represents same or similar element or the element with same or like function.Retouched below with reference to accompanying drawing The embodiment stated is exemplary, is only used for explaining the present invention, and is not considered as limiting the invention.

One has the extensive mimo system modeling of M foundation station antenna and K root user antennas as shown in Figure 1.Use H_DLRepresent Down physical channel gain (between antenna for base station to terminal antenna) of Equivalent Base-Band, then after having cascaded RF receiving and transmission module Comprehensive down channel and up channel it is as follows：

Wherein T and R represent the radiofrequency emitting module of antenna for base station and the Equivalent Base-Band gain matrix of Receiver Module, WithRepresent the radiofrequency emitting module of user antenna and the Equivalent Base-Band gain matrix of Receiver Module.

It is easy to obtain

Wherein C_A=RT^-1,Ignore the coupling between different antennae, T, R,All it is diagonal matrix.

T=diag (t₁,t₂...,t_M),

R=diag (r₁,r₂...,r_M)

For relative calibration, it is only necessary to obtain C_A=RT^-1=diag (c₁,,...,c_M),c_i=r_i/t_i(i=1,2 ..., M)。

LS algorithms are precision one kind higher in known algorithm, and it includes three below step：

(1) the two-way cascaded channel gain between any two foundation station antenna is obtained

(2) structural matrix W, and the characteristic vector corresponding to Eigenvalues Decomposition acquisition minimal eigenvalue is carried out, as calibrate Vectorial c=(c₁,c₂,...,c_M)^T.Wherein W is：

LS algorithms calibration complexity is too high, and antenna for base station is fifty-fifty divided into m groups by the present invention as far as possible, then to every group Inner antenna carries out part LS calibration, then calibration carrying out group to m groups antenna, by taking m=4 as an example, is grouped scheme as shown in Figure 2.

Calibration in group can easily pass through LS algorithms to realize.However, the calibration program between group has many kinds, it is considered to It is single to calibrate (as shown in Figure 3) two schemes with reference to group calibration (as shown in Figure 2) and adjacent sets recursion.

Corresponding to single with reference to group local alignment, respectively to 4 antenna sets B₁,B₂,B₃,B₄LS calibrations are carried out, local school is obtained Quasi- vectorLS calibrations are carried out to the entirety that all antennas are constituted, global alignment vector is labeled as c=(c₁, c₂,c₃,c₄)^T, wherein c_i(i=1,2,3,4) antenna sets B is corresponded in the global alignment vector of expression_iSubvector.

Note B_ijTo carry out B_iAnd B_jThe reference antenna set being related to is calibrated between group, to B_ijLS calibrations are also carried out, accordingly Alignment vector is designated as

Define B_ijk=B_ij∩B_k, k=i, j

NoteForIn correspond to B_ijkSubvector,ForIn correspond to B_ijkSubvector, c_ijkIt is c_kMiddle correspondence In B_ijkSubvector.

The local alignment vector obtained with LS algorithms is that corresponding overall situation alignment vector component is multiplied by constant value (this Constant value is different to different antenna sets).If not having noisy influence in calibration process,WithAll it is c_ijk Multiple, in the presence of noise, following expression formula can be obtained：

Define d_ijk=α_ijk/β_ijk, it is apparent to d_ijkEstimation can be by total least square (Total Least Square) algorithm is addressed.DefinitionIt is d_ijkEstimate

OrderIt is the estimate of global alignment vector c, then can obtains global by local alignment vector The estimate of alignment vector

For the calibration of adjacent sets recursion, pattern is as shown in Figure 3.Correspondingly, can be obtained by local alignment vector

Fig. 4 shows the MSE of bank calibration scheme.It is single complexity to be dropped with reference to group calibration and the calibration of adjacent sets recursion It is low to O (M³/16).However, due to the reduction of path loss, the accuracy of adjacent sets recursion calibration program is higher.

The calibration of adjacent sets recursion is expanded into two-dimensional array scene, as shown in figure 5, the performance after being calibrated using local recursion As shown in Figure 6.Understand, the adjacent sets recursion calibration accuracy loss very little based on local alignment, but complexity is reduced to O (M³/ 16), Calibration speed is greatly improved.

Claims (6)

1. a kind of extensive MIMO reciprocities calibration method based on local alignment, it is adaptable to the extensive MIMO systems of time division duplex System, it is characterised in that comprise the following steps：

All antenna for base station are divided into different groups by step A, and each group of base stations antenna is calibrated using LS algorithms, obtain each The local alignment coefficient of group antenna for base station；

Step B, to different groups, selects some foundation station antennas as reference antenna, obtains calibration factor between different groups of group；

Step C, using calibration factor between the local alignment coefficient and group obtained in step A and B, is calculated global calibration factor C=(c₁,c₂,...,c_M)^T, c_i=r_i/t_i(i=1,2 ..., M), M is the sum of antenna for base station, t_iIt is penetrating for antenna for base station i The Equivalent Base-Band gain of frequency transmitter module, r_iIt is the Equivalent Base-Band gain of the Receiver Module of antenna for base station i；

Step D, using global calibration factor, the transmitting and receiving module to antenna for base station is adjusted, and realizes relative calibration process.

2. the extensive MIMO reciprocities calibration method of local alignment is based on according to claim 1, it is characterised in that described In step A, antenna for base station is fifty-fifty divided into m groups as far as possible, while setting threshold value d₀, it is ensured that between every group of inside antenna Distance is less than d₀。

3. the extensive MIMO reciprocities calibration method of local alignment is based on according to claim 1, it is characterised in that described It is as follows the step of calibrated using LS algorithms in step A：

Step 1, in antenna sets, mutually sends pilot tone between any two foundation station antenna, measure any two foundation station antenna Between channel gain

Step 2, using the channel gain between any two foundation station antennaStructural matrix W：

$\begin{array}{cc}W=\[w_{i,j}\]& w_{i,j}=\left\{\begin{array}{cc}\underset{j:(i,j)\∈F}{\Σ}|{\tilde{h}}_{i\→j}{|}^2& forj=i\\ -{\tilde{h}}_{j\→i}{\tilde{h}}_{i\→j}^{*}& forj\≠i\end{array}\right.\end{array},$

Step 3, Eigenvalues Decomposition is carried out to matrix W, obtains the characteristic vector corresponding to its minimal eigenvalue, as local alignment CoefficientWherein i and j is the sequence number of antenna for base station, and m is the sequence number of antenna sets, M_mIt is m group antennas Number.

4. the extensive MIMO reciprocities calibration method of local alignment is based on according to claim 1, it is characterised in that described In step B, for antenna sets m₁With antenna sets m₂When being calibrated between group, it is assumed that choose n roots antenna altogether as reference antenna, then from Antenna sets m₁Middle selected distance antenna sets m₂Nearest n/2 root antennas, from antenna sets m₂Middle selected distance antenna sets m₁Nearest n/ 2 antennas, constitute a new antenna sets

5. the extensive MIMO reciprocities calibration method of local alignment is based on according to claim 1, it is characterised in that described In step B, when being calibrated using adjacent antenna group recursion, some base station antenna compositions are only chosen from the antenna sets of adjacent antenna One new new antenna setsNew antenna sets are carried out with LS calibrations described in claim 3, the group of different antennae group is obtained Between calibration factor

6. the extensive MIMO reciprocities calibration method of local alignment is based on according to claim 1, it is characterised in that described In step B, during using single calibration with reference to group, antenna sets m is taken_n/2As reference antenna group, some are chosen from other antenna sets Antenna for base station and antenna sets m_n/2In some base station antenna sets into the new antenna sets of antennaNew antenna sets are carried out LS calibrations described in claim 3, obtain calibration factor between the group of different antennae group