CN105356917A - Interference suppression transmission method in large-scale MIMO (Multiple-Input Multiple-Output) heterogeneous network - Google Patents

Abstract

The invention provides an interference suppression transmission method in a large-scale MIMO (Multiple-Input Multiple-Output) heterogeneous network, and belongs to the technical field of wireless communication. The interference suppression transmission method in the large-scale MIMO heterogeneous network comprises the steps of firstly judging interference from a macro-base station to a micro-base station is strong interference or weak interference; and then performing different treatments by the micro-base station to further obtain a pre-coding matrix of the micro-base station and enabling a data rate to be maximum, wherein when the interference is weak interference, the interference from the macro-base station is processed as noise; and when the interference is strong interference, interference deletion processing is adopted. According to the interference suppression transmission method in the large-scale MIMO heterogeneous network, the existing problems of great pre-coding design calculation amount and low efficiency are solved, the calculation efficiency is increased as the solution of the pre-coding matrix can be obtained only by calculating for one time, the system processing time is shortened, and significance for increasing a communication system efficiency is obtained.

Description

Interference suppression transmission method in large-scale MIMO heterogeneous network

Technical Field

The invention belongs to the technical field of wireless communication, and relates to a transmission method for suppressing inter-user interference and cross-layer interference in a large-scale MIMO (Multiple-input Multiple-Output) heterogeneous network scene.

Background

MIMO is an abbreviation for multiple input multiple output (mlmo). The MIMO technology can implement diversity and space division multiplexing of transmission/reception by providing multiple antennas, thereby providing diversity gain and multiplexing gain, and can improve system capacity by multiples without increasing spectrum resources and antenna transmission power. Nowadays, MIMO technology has been widely applied in the field of wireless communication, for example, in current 4G communication, both multi-antenna transmit diversity and multi-stream space division multiplexing are applied. The popularization of MIMO technology makes space a resource that can be used to improve performance and can increase the coverage of a wireless system.

With the continuous popularization and development of multimedia terminals, the demand of users for communication efficiency is increasing, the demand for the number of antennas is also increasing, and large-scale MIMO is coming up. The large-scale MIMO system has a very large array element number, can provide higher diversity gain and multiplexing gain, and can meet the requirements of higher and higher data rates.

Meanwhile, in order to ensure the service quality of each user, the cell also presents the characteristic of isomerization. However, in the heterogeneous network, the problem of cell coverage occurs, and besides the inter-user interference, the cross-layer interference is also not negligible. One common approach to suppress interference is also by designing the precoding to meet the system requirements. However, the traditional method usually adopts an iterative algorithm, so that the calculation amount is large and the efficiency is low. In massive MIMO, there has been a lot of work related. Document [1] considers a two-layer precoding system, and proposes a method for obtaining a precoding matrix through iteration. Document [2] combines outer layer coding with a channel, thereby reducing the dimension of the channel and reducing the complexity of channel estimation. Meanwhile, it is assumed that the range of user activity in the cluster is small, so the inner layer code is updated at short time intervals, and the outer layer code is updated at long time intervals. The outer precoding design is a combination of zero-forcing (ZF) precoding and matched filter, and the weights of ZF and matched filter are assigned to control their respective weights. The inner layer coding is used to suppress residual interference. Document [3] considers a more complex situation, combining MIMO and Orthogonal Frequency Division Multiplexing (OFDM), and performing IFFT (inverse fast fourier transform) and FFT (fast fourier transform) processing on the transmitting side and the receiving side, respectively. The outer layer coding uses a pre-quantized codebook (the codebook is an array cluster vector of direction angles, and the direction angles are quantized according to equal angle spacing or equal sine value spacing), and elements of beam pointing paths are selected from the codebook to be used as the outer layer coding. The peak-to-average power ratio of the system in the case of known codes with normalized characteristics, etc. are discussed. Document [4] considers two-layer precoding under large-scale MIMO, a receiving end has a single antenna, users are clustered, outer-layer precoding is designed through channel statistical information to suppress inter-cluster interference, and inner-layer precoding suppresses inter-user interference. And, when the channel satisfies a certain condition, the outer layer precoding elements are modulo equal.

There is also some work in the suppression of inter-cell interference. Document [5] uses a "partial frequency reuse" method, in which cell center users use the same frequency resources, and cell edge users use orthogonal frequency resources, so that interference can be suppressed. Document [6] uses a time domain orthogonal method, that is, a macro base station sends some blank frames in a time range, thereby mitigating cross-layer interference to a micro cell. Both document [7] and document [8] use a non-linear method to suppress interference. Document [7] considers an interference suppression method using active interference cancellation in a mixed interference scenario, and obtains precoding through iteration. Document [8] divides interference scenes in a heterogeneous network into three types (weak, strong and weak) according to interference strength, and a user selects a transmission scheme for deleting interference as noise or active interference in different scenes, so that a precoding analysis solution in each scene is provided.

Meanwhile, some large-scale antennas are applied to related work in heterogeneous network scenes. Document [9] proposes a scene in which a macro base station in a heterogeneous network is provided with a large-scale antenna and coexists with a plurality of single-antenna micro cells, analyzes system performance, and provides a curve of micro cell spatial spectrum efficiency with respect to micro cell density. Document [10] considers suppression of cross-layer interference in a large-scale antenna heterogeneous network, users are firstly grouped, and different cooperations are adopted between macro cells and micro cells to manage the cross-layer interference.

Although interference management measures of large-scale antennas and heterogeneous networks are provided from different angles in the existing work, precoding analysis solutions are not provided in the large-scale MIMO heterogeneous networks aiming at different interference scenes such as strong interference or weak interference.

The references [1] to [10] are specifically as follows:

[1]AlkhateebA,AyachOE,LeusG,etal.Hybridprecodingformillimeterwavecellularsystemswithpartialchannelknowledge[C]//InformationTheoryandApplicationsWorkshop(ITA).SanDiego,CA:IEEE,2013:1-5；

[2]ChenJunting,andLauVKN.Two-TierPrecodingforFDDMulti-cellMassiveMIMOTime-VaryingInterferenceNetworks[J].IEEEJournalonSelectedAreasinCommunications,2014,32(6):1230-1238；

[3]KimC,SonJS,KimT,etal.OntheHybridBeamformingwithSharedArrayAntennaformmWaveMIMO-OFDMSystems[C]//WirelessCommunicationsandNetworkingConference(WCNC).Istanbul:IEEE,2014:335-340；

[4]AdhikaryA,NamJ,AhnJY,etal.JointSpatialDivisionandMultiplexing—TheLarge-ScaleArrayRegime[J].IEEETransactionsonInformationTheory,2013,59(10):6441-6463；

[5]BoudreauG,PanickerJ,GuoN,etal.Interferencecoordinationandcancellationfor4Gnetworks[J].IEEECommunicationsMagazine,2009,47(4):74-81；

[6]Lopez-PerezD,GuvencI,G.DLR,etal.EnhancedInter-CellInterferenceCoordinationChallengesinHeterogeneousNetworks[J].IEEEWirelessCommunications,2011,18(3):22-30；

[7]WangYunlu,TianYafei,LiYang,etal.Coordinatedprecodingandproactiveinterferencecancellationinmixedinterferencescenarios[C]//WirelessCommunicationsandNetworkingConference(WCNC).Istanbul:IEEE,2014:554-558；

[8]LiYang,TianYafei,andYangChenyang.BeamformingDesignwithProactiveInterferenceCancellationinMISOInterferenceChannels[J].EURASIPJournalonAdvancesinSignalProcessing,2015:67(2Aug.2015)；

[9]KountourisM,PappasN.HetNetsandmassiveMIMO:Modeling,potentialgains,andperformanceanalysis[C]//AntennasandPropagationinWirelessCommunications(APWC).Torino:IEEE,2013:1319-1322；

[10]AdhikaryA,DhillonHS,CaireG.Massive-MIMOMeetsHetNet-InterferenceCoordinationThroughSpatialBlanking[J].IEEEJournalonSelectedAreasinCommunications,2015,33(6):1171-1186。

disclosure of Invention

Aiming at the problems of large calculation amount and low efficiency of the existing precoding design, the invention provides an interference suppression transmission method in a large-scale MIMO heterogeneous network in order to suppress interference between users and cross-layer interference. Aiming at the condition that a macro cell and a micro cell are overlapped with each other, the macro base station is provided with a large-scale antenna, and the micro user adopts different interference processing modes according to different intensities of interference generated by the macro base station: and active interference deletion is carried out when the interference is strong, and the interference is taken as noise when the interference is weak. Under different processing modes, the micro base station precoding gives an analytic solution.

The invention relates to an interference suppression transmission method in a large-scale MIMO heterogeneous network, which comprises the following steps:

step one, judging whether the interference intensity of a macro base station on a micro base station is strong interference or weak interference;

step two, the micro base station adopts a corresponding processing mode according to the interference intensity of the macro base station: (1) when the interference is weak interference, the interference of the macro base station is taken as noise processing, and the step 2.1 is executed; (2) and when the interference is strong interference, the interference deletion processing is adopted, and the step 2.2 is executed.

In the second step, H is set_mmChannel vector, V, for macro base station to macro user of the g-th cluster_mOuter precoding matrix or complete precoding matrix for macro base station to macro user data of the g cluster, H_ppFor the channel from the g-th micro base station to the g-th micro user, V_pPrecoding matrix for the g-th micro base station, H_pmA channel from the macro base station to the g-th micro user; p_mNormalized power for macro base station transmit power and noise power,P_pnormalized power for the micro base station transmit power and noise power,P_macrotransmitting power, P, for macro base stations_picoFor transmitting power, P, to the micro base station_NIs the noise power; g is a positive integer.

Step 2.1, the treatment mode (1) is as follows: data rate R of micro-users_picoExpressed as:

$R_{pico}=\log |I+P_p{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H{(I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H)}^{-1}|,$

wherein the upper corner H represents transpose, I is an identity matrix,to be a regular matrix, the regular matrix is decomposed and represented as:Q₁a middle matrix for decomposing the left normal matrix;

(1.1) when a micro-user transmits a single stream data, V_pIs composed ofMaximum eigenvalue λ of_maxA corresponding feature vector x;

(1.2) when the micro-user transmits n stream data, n is an integer greater than 1, obtainingN maximum eigenvalues and arranged in descending order of λ₁,λ₂,…λ_n，x₁,x₂,…x_nAre each lambda₁,λ₂,…λ_nCorresponding eigenvector to obtain a precoding matrix V_p＝[x₁x₂…x_n]。

Step 2.2, the treatment mode (2) is as follows: and the data rate R is expressed as:

$R=log|I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H|+log|I+P_p{(I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H)}^{-1}{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H)|;$

wherein, the upper corner mark H represents transposition, I is an identity matrix, and the normal matrix isIs decomposed intoQ₂Is an intermediate matrix obtained by decomposition;

(2.1) when the micro-user transmits single stream data, set V_pIs composed ofMaximum eigenvalue λ of_maxA corresponding feature vector x;

(2.2) when the micro-user transmits n stream data, n is an integer greater than 1, selectingOrthogonal eigenvector x corresponding to the largest n eigenvalues₁,x₂,…x_nForming a precoding matrix V_p，V_p＝[x₁x₂…x_n]。

The invention has the advantages and positive effects that: compared with the traditional method, the interference suppression transmission method provided by the invention does not need to carry out iterative operation, can obtain the solution of the precoding matrix by only carrying out one-time calculation, improves the calculation efficiency, and shortens the system processing time, so that the interference suppression transmission method provided by the invention is very significant for improving the efficiency of a communication system.

Drawings

Fig. 1 is a schematic view of a scenario in which a macro cell and a micro cell overlap each other;

fig. 2 is a schematic diagram of an interference suppression transmission method in a massive MIMO heterogeneous network according to the present invention;

fig. 3 is a diagram of performance curves for different transmission methods.

Detailed Description

The present invention will be described in further detail with reference to the accompanying drawings and examples.

The interference suppression transmission method in the large-scale MIMO heterogeneous network aims at the condition that a macro cell and a micro cell are mutually overlapped, and a large-scale antenna is arranged on a macro base station to perform cooperation suppression on interference. The macro base station and the micro base station share channel information and precoding information, and the micro base station calculates the precoding of the receiving end according to the channel information and the precoding information of the macro base station. The macro base station may adopt any precoding mode, such as single-layer precoding and two-layer precoding, and the outer layer may be analog precoding or digital precoding when the two-layer precoding is performed. The micro base station adopts different processing methods according to different interference strengths of the macro base station: and when the interference is strong, interference deletion is adopted, and when the interference is weak, the interference of the macro base station is taken as noise.

In a massive MIMO heterogeneous network scenario, consider a case where a macro cell and a micro cell overlap each other, as shown in fig. 1.

A macro base station is MBS, a macro user is Muser, a micro base station is PBS, and a micro user is Puser. The macro users are clustered, a micro cell is arranged near each macro user, the micro cell near the g-th cluster macro user in the macro cell is defined as the g-th micro cell, and the micro base station and the micro user in the corresponding micro cell are respectively the g-th micro base station and the g-th micro user.

The number of macro base station antennas is M_macroThe number of antennas of the g-th micro base station is M_pico-BS-gThe number of the g-th micro-user antenna is M_pico-User-g. Consider that in the scenario of a heterogeneous network, a macro cell and a micro cell operate in the same frequency band. The macro cell is a large-scale MIMO system, the macro base station is provided with large-scale antennas, macro users are arranged into G clusters, the users have multiple antennas, each user receives multi-stream data, and each cluster receives K data in common_gThe macro base station transmits S stream data together; each micro cell serves 1 micro user, the g micro user receives N_gStreaming data. Macro base station transmitting power is P_macroThe micro base station has a transmission power of P_picoNoise power of P_N. Normalized power of macro base station transmitting power and noise power isNormalized power of micro base station transmitting power and noise power is

Let the number of antennas of the ith macro user be M_iIts received signal y_iCan be expressed as:

$y_i=P_m{H}_{i,m}{V}_{i,m}{x}_{i,m}+P_m\underset{k\≠i}{\Σ}{H}_{i,m}{V}_{k,m}{x}_{k,m}+P_p{H}_{ip}{V}_p{x}_p+n_i$

H_i,mfor the channel between the macro base station and the ith macro user,wherein,represents M_i×M_macroDimension of complex vector space, V_i,mA precoding matrix for the ith macro user data,S_ithe number of data streams received for the ith macro user,represents M_macro× S-dimensional complex vector space, x_i,mTransmission data indicating the ith macro user;for the interference, V, produced by the other macro-users to which the ith macro-user is subjected_k,mPrecoding matrix for kth macro user data, x_k,mTransmission data indicating the kth macro user; h_ipV_px_pIndicating interference from the micro-user to the ith macro-user, H_ipFor the interference channel from the micro base station to the i-th macro user, V_pPrecoding matrix, x, for micro base stations_pTransmitting data for the micro base station; n is_iRepresenting the noise of the ith macro-user,representing a distribution obeying a mean of 0 and a variance of I. Wherein I represents an identity matrix, that is, n_iThe autocorrelation value of (2) is 1, and the random generation is independent of each other. The channel between the base station and the user in the method of the invention is the downlink channel.

Since macro users in each cluster are geographically distant, interference between different micro cells and interference to micro users by non-nearby clusters is negligible. Thus, the g-th micro-user receives the signal y_gCan be expressed as:

y_g＝P_pH_g,pV_g,px_g,p+P_mH_gmV_mx_m+n_p

H_g,pfor the channel between the g-th micro-user and the micro base station serving it,represents M_pico-User-g×M_pico-BS-gDimension of complex vector space, V_g,pPrecoding matrix, x, representing the g-th micro-user data_g,pFor the data transmitted by the micro base station,represents N_gA complex vector space is maintained and,for the interference channel from the macro base station to the micro user,represents M_pico-User-g×M_macroA complex vector space is maintained and,for the data sent by the macro base station,representing an S-dimensional complex vector space, P_mH_pmV_mx_mFor the interference from the macro base station to which the user is subjected, n_pRepresenting the noise of the micro-users, the invention makes the noise suffered by the micro-users uniformly n_pTo indicate that the user is not in a normal position,is noise. H_pmRepresenting the channel from the micro base station to the macro user, V_mRepresenting a precoding matrix, x, of a macro base station_mIndicating transmission data of the macro base station.

The macro base station may employ single layer precoding or two layer precoding, and the outer layer of the two layer precoding may be analog or digital precoding. It should be noted that both macro and micro users need to feed back channel information to the base station, and at any time, the micro base station needs to receive the channel information of the macro cell from the macro base station. Meanwhile, when the macro base station adopts single-layer precoding, the micro base station needs to utilize all precoding information of the macro base station. When the macro base station adopts two layers of precoding, the micro base station only needs to utilize the information of the outer layer precoding of the macro base station. Both of these messages require the macro base station to send to the micro base station.

According to different interferences generated by the macro base station to the micro base station, the micro base station adopts different precoding schemes under different signal to interference and noise ratio scenes, namely: in the scene (1), a micro user is subjected to weak interference, and the interference is taken as noise to directly solve received data; and (3) the micro-user in the scene (2) is subjected to strong interference, and the influence of the interference is eliminated by adopting an active interference deleting method. In different scenes, the precoding form of the micro base station is different.

Because the spatial positions of clusters of macro users are different and the correlation between channels of different clusters of users is poor, when the micro base station performs cross-layer interference suppression, the g-th micro user is only interfered by the signal of the g-th cluster of macro users.

First, the definition of the character concerned will be explained: h_mmChannel, V, for macro base station to macro user of the g-th cluster_mOuter precoding matrix or complete precoding matrix for macro base station to macro user data of the g cluster, H_ppFor the channel from the g-th micro base station to the g-th micro user, V_pPrecoding matrix for the g-th micro base station, H_pmChannel from macro base station to the g-th micro-user, H_mpThe channel from the g-th micro base station to the g-th cluster macro user.

The interference suppression transmission method in the massive MIMO heterogeneous network mainly comprises two steps, as shown in FIG. 2: firstly, judging the interference intensity generated by a macro base station; and step two, the micro base station adopts corresponding different processing modes according to different interference strengths of the macro base station: and when the interference is strong interference, the interference is deleted, and when the interference is weak interference, the interference of the macro base station is regarded as noise.

Firstly, the following 2 methods can be adopted for judging the interference intensity generated by the macro base station:

(1) the data rates corresponding to the two processing modes are calculated in advance at the micro base station side: if the data rate of the interference as the noise is higher than the data rate of the interference deletion, the interference strength of the macro base station is considered to be weak interference, and the interference is considered to be the noise; otherwise, the interference intensity of the macro base station is considered to be strong interference, and interference deletion is carried out.

(2) Calculating the 2 norm of the channel matrix at the receiving end of the micro-user: if the 2 norm of the channel matrix of the direct transmission link from the micro base station to the micro user is greater than the 2 norm of the channel matrix of the cross link from the macro base station to the micro user, the weak interference scene is considered, and the interference is directly considered as noise; otherwise, the scene is regarded as a strong interference scene, and interference deletion is carried out.

When the micro base station adopts different receiving end processing modes, the precoding matrixes of the micro base station are different, and the precoding is calculated according to the precoding information of the macro base station and the channel information of the macro cell and the micro cell to obtain an analytic solution.

The following describes the implementation process of calculating and obtaining the precoding of the micro base station to maximize the data rate in the way of processing the macro base station interference as noise and performing interference cancellation.

And 2.1, the micro user processes the macro base station interference as noise.

Consider first the case where a micro-user transmits single stream data. When the interference of the macro base station to the micro user is weak, the micro user can regard the interference generated by the macro base station as noise, and at the moment, the data rate R of the micro user_picoComprises the following steps:

$R_{pico}=log|I+\frac{P_p{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H}{I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H}|,$

namely, it is

$R_{pico}=log|I+P_p{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H{(I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H)}^{-1}|,$

Wherein the upper corner H represents transpose, I is an identity matrix,for a regular matrix, the regular matrix is decomposed and can be written as:

${(I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H)}^{-1}=Q_1^H{Q}_1,$

wherein Q is₁To decompose the middle matrix of the left-side normal matrix,is a matrix Q₁The transposed matrix of (2).

Thus, it is possible to obtain:

$R_{pico}=log|I+P_p{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H{Q}_1^H{Q}_1|,$

using the matrix property of | I + BC | ═ I + CB |, we can obtain:

$R_{pico}=log|I+P_p{Q}_1{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H{Q}_1^H|,$

will be provided withWritten form of singular value decomposition $I+P_p{Q}_1{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H{Q}_1^H={\mathrm{UAU}}^H,$ Wherein U is a matrixA is a diagonal element matrixA matrix of singular values of; then there is

$R_{\mathrm{pico}}=\log \left|{\mathrm{UAU}}^H\right|=\log \left|U\right|\left|A\right|\left|U^H\right|=\log \left|A\right|=\log \left|\begin{array}{cccc}{K}^2+1& 0& 0& 0\\ 0& 1& 0& 0\\ 0& 0& 1& 0\\ 0& 0& 0& 1\end{array}\right|=\log (K^2+1),$

Intermediate matrix $K=\left|\right|\sqrt{P_p}{Q}_1{H}_{pp}{V}_p|{|}_2.$

Setting V_pIs composed ofMaximum eigenvalue λ of_maxCorresponding feature vector x, when R_picoObtaining a maximum value R_picomaxIs composed of

R_picomax＝log(P_pλ_max+1)

When the micro-user transmits n stream data, n is an integer greater than 1, then selectingMaximum n eigenvalues λ₁,λ₂,…λ_nCorresponding orthogonal eigenvectors x₁,x₂,…x_nAs a beamforming vector for each stream data, the obtained precoding matrix is:

V_p＝[x₁x₂…x_n],

at this time, R_picoObtaining a maximum value R_picomaxIs composed of

$R_{picomax}=\underset{k=1}{\overset{n}{\Σ}}log(P_p{\mathrm{\λ}}_k+1)$

Wherein λ is₁,λ₂,…λ_nArranged in descending order, x₁,x₂,…x_nAre each lambda₁,λ₂,…λ_nThe corresponding feature vector.

And 2.2, actively deleting interference by the micro-user.

The invention solves the interference as a signal and then subtracts the interference from the useful signal. Theoretically, the interference can be solved without error as long as the transmission rate of the interference is lower than the channel capacity when the interference is considered as a received signal. Therefore, it is sufficient that no interference item is directly written in the data rate item indicating the interference deletion in the following equation (since the transmission rate of the interference is specified to be smaller than the capacity in the condition).

Also, first, consider the case where a micro-user transmits single-stream data. When the micro user performs active interference cancellation to suppress interference of the macro base station, the sum data rate R formula is expressed as:

R＝R_macro+R_pico,

R_macrorepresenting the data rate of the macro base station.

In order to ensure that the data of the macro base station can be solved in the direct transmission link and the cross link, the data rate of the macro base station meets the following conditions:

$\left\{\begin{array}{c}{R}_{macro}\≤\log |I+\frac{P_m{H}_{mm}{V}_m{V}_m^H{H}_{mm}^H}{I+P_p{H}_{mp}{V}_p{V}_p^H{H}_{mp}^H}|,\\ R_{\mathrm{ma}cro}\≤\log |I+\frac{P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H}{I+P_p{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H}|,\end{array}\right.$

thus, there are

$R_{macro}=\log \min \left(\right|I+\frac{P_m{H}_{mm}{V}_m{V}_m^H{H}_{mm}^H}{I+P_p{H}_{mp}{V}_p{V}_p^H{H}_{mp}^H}|,|I+\frac{P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H}{I+P_p{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H}\left|\right).$

Since the micro base station is closer to the micro user than the macro base station, when the cross link interference is solved, the signal transmitted to the micro user by the micro base station is taken as the interference, and at this time, the data of the micro base station forms strong interference. Therefore, the present invention makes an assumption that the channel environment of the cross link is considered worse. Thus, the macro user data rate formula can be written as

$R_{macro}=log|I+\frac{P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H}{I+P_p{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H}|,$

And due to

$R_{pico}=log|I+P_p{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H|,$

Thus, there are

$R=R_{macro}+R_{pico}=log|I+P_p{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H|,$

Further transforming the formula to obtain

$\begin{array}{c}R=\log \left|(I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H)(I+P_p{\left(I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H\right)}^{-1}{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H)\right|\\ \begin{array}{c}=\log \\ =\log \end{array}\left|\begin{array}{c}I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H\\ I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H\end{array}\right|\begin{array}{c}|I+P_p{(I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H)}^{-1}{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H|\\ +\log |I+P_p{(I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H)}^{-1}{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H|\end{array}\end{array},$

In the above formula, the first and second carbon atoms are,may be considered constant.

Also, the normal matrixDecomposition is performed, expressed as:

${(I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H)}^{-1}=Q_2^H{Q}_2,$

wherein Q is₂The intermediate matrix resulting from the decomposition of the left-side normal matrix,is a matrix Q₂The transposed matrix of (2).

Then R is transformed into:

$\begin{array}{c}R=\log |I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H|+\log |I+P_p{Q}_2^H{Q}_2{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H|\\ =\log |I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H|+\log |I+P_p{Q}_2{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H{Q}_2^H)|\end{array},$

setting V_pIs composed ofMaximum eigenvalue λ of_maxCorresponding feature vector x, when R takes the maximum value R_maxThe following are:

$R_{max}=log|I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H|+log(P_p{\mathrm{\λ}}_{max}+1).$

similarly, if the micro-user transmits n-stream data, the micro-user selectsOrthogonal eigenvector x corresponding to the largest n eigenvalues₁,x₂,…x_nAs a beamforming vector for each stream data, i.e. a precoding matrix V_p＝[x₁x₂…x_n]At this time, R takes the maximum value R_maxComprises the following steps:

$R_{max}=log|I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H|+\underset{k=1}{\overset{n}{\Σ}}log(P_p{\mathrm{\λ}}_k+1).$

wherein λ is₁,λ₂,…λ_nArranged in descending order, x₁,x₂,…x_nAre each lambda₁,λ₂,…λ_nThe corresponding feature vector.

The method of the present invention is subjected to simulation testing as follows. The simulation parameters are selected as follows: the number of macro base station antennas is 128, the radius of a macro cell is 500m, the macro base station antennas are divided into 4 clusters, and the total number of user antennas in each cluster is 4; each micro base station serves 1 micro user, and the number of the micro user antennas is 4. The transmitting power of the macro base station is 46dBm, the transmitting power of the micro base station is 30dBm, and the signal-to-noise ratio of the edge of the macro cell is 5 dB. The channel of the macro cell adopts a 3GPP channel model, the small-scale fading of the micro cell channel obeys Rayleigh distribution, and the large-scale fading meets the definition of the 3GPP channel model. The channel statistical information is obtained through statistics after 1000 times of channel generation, and the channel instantaneous information is realized through single generation of the channel under the same condition. The results were averaged over 1000 replicates.

The simulation results are shown in fig. 3. MRT and eICIC methods are selected for comparison with the method of the invention. The MRT method is to use MRT precoding (maximum ratio transmit precoding) by the micro base station, and to treat all interference as noise. In an elcic (enhanced inter-cell interference coordination) method, a macro cell is orthogonal in time to a micro cell. The curves eICIC _2, eICIC _3, and eICIC _4 correspond to data rates when the number of occupied time slots of the macro cell is 2,3, and 4, respectively, and the total number of downlink time slots in each data frame is considered to be 6. It should be noted that as the number of time slots allocated to the macro cell increases, the system throughput decreases, because the micro cell has a better channel condition and the system throughput is mainly provided by the micro cell. As can be seen from simulation results, the system and the data rate of the scheme provided by the application are obviously superior to those of the MRT method and the eICIC method.

Claims (5)

1. An interference suppression transmission method in a large-scale MIMO heterogeneous network is characterized by comprising the following implementation steps:

step one, judging whether the interference intensity of a macro base station on a micro base station is strong interference or weak interference;

step two, the micro base station adopts a corresponding processing mode according to the interference intensity of the macro base station: (1) when the interference is weak interference, the interference of the macro base station is taken as noise processing, (2) when the interference is strong interference, the interference deletion processing is adopted;

let H_mmChannel direction from macro base station to macro user in the g-th clusterAmount, V_mOuter precoding matrix or complete precoding matrix for macro base station to macro user data of the g cluster, H_ppFor the channel from the g-th micro base station to the g-th micro user, V_pPrecoding matrix for the g-th micro base station, H_pmA channel from the macro base station to the g-th micro user; p_mNormalized power for macro base station transmit power and noise power,P_pnormalized power for the micro base station transmit power and noise power,P_macrotransmitting power, P, for macro base stations_picoFor transmitting power, P, to the micro base station_NIs the noise power; g is a positive integer;

step 2.1, the treatment mode (1) is as follows: data rate R of micro-users_picoExpressed as:

$R_{pico}=log|I+P_p{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H{(I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H)}^{-1}|,$

wherein the upper corner H represents transpose, I is an identity matrix,to be a regular matrix, the regular matrix is decomposed and represented as:Q₁a middle matrix for decomposing the left normal matrix;

(1.1) when a micro-user transmits a single stream data, V_pIs composed ofMaximum eigenvalue λ of_maxA corresponding feature vector x;

(1.2) when the micro-user transmits n stream data, n is an integer greater than 1, obtainingN maximum eigenvalues and arranged in descending order of λ₁,λ₂,…λ_n，x₁,x₂,…x_nAre each lambda₁,λ₂,…λ_nCorresponding eigenvector to obtain a precoding matrix V_p＝[x₁x₂…x_n]；

Step 2.2, the treatment mode (2) is as follows: and the data rate R is expressed as:

$R=log|I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H|+log|I+P_p{(I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H)}^{-1}{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H)|;$

wherein, the upper corner mark H represents transposition, I is an identity matrix, and the normal matrix isIs decomposed intoQ₂Is an intermediate matrix obtained by decomposition;

(2.1) when the micro-user transmits single stream data, set V_pIs composed ofMaximum eigenvalue λ of_maxA corresponding feature vector x;

(2.2) when the micro-user transmits n stream data, n is an integer greater than 1, selectingOrthogonal eigenvector x corresponding to the largest n eigenvalues₁,x₂,…x_nForming a precoding matrix V_p，V_p＝[x₁x₂…x_n]。

2. The method according to claim 1, wherein in the first step, the method for determining the interference strength generated by the macro base station comprises: calculating the 2 norm of the channel matrix at the receiving end of the micro-user: if the 2 norm of the channel matrix of the direct transmission link from the micro base station to the micro user is greater than the 2 norm of the channel matrix of the cross link from the macro base station to the micro user, the interference strength of the macro base station is weak interference; otherwise, the interference strength of the macro base station is strong interference.

3. The method according to claim 1, wherein in the first step, the method for determining the interference strength generated by the macro base station comprises: and D, calculating the data rates corresponding to the two processing modes in the step two in advance at the micro base station end: if the data rate of the interference as noise is higher than the data rate of the interference deletion, the interference strength of the macro base station is weak interference; otherwise, the interference strength of the macro base station is strong interference.

4. The method as claimed in claim 1, wherein in step 2.1, the interference generated by the macro base station is regarded as noise by the micro user, and the interference is considered as noise by the micro user

$R_{pico}=log|I+\frac{P_p{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H}{I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H}|,$

Namely, it is

$R_{pico}=log|I+P_p{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H{(I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H)}^{-1}|,$

Will be regular matrixIs decomposed into

Thus obtaining $R_{pico}=log|I+P_p{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H{Q}_1^H{Q}_1|,$ Further obtaining:

$R_{pico}=log|I+P_p{Q}_1{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H{Q}_1^H|,$

will be provided withWritten form of singular value decomposition $I+P_p{Q}_1{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H{Q}_1^H={\mathrm{UAU}}^H,$ Then there is

$R_{pico}=log\left|{\mathrm{UAU}}^H\right|=log\left|U\right|\left|A\right|\left|U^H\right|=log\left|A\right|=log\left|\begin{array}{cccc}{K}^2+1& 0& 0& 0\\ 0& 1& 0& 0\\ 0& 0& 1& 0\\ 0& 0& 0& 1\end{array}\right|=log(K^2+1),$

Wherein U and A are matrices respectivelyThe eigenvector matrix and the matrix with diagonal elements as singular values; intermediate matrix $K=\left|\right|\sqrt{P_p}{Q}_1{H}_{pp}{V}_p|{|}_2;$

If the micro-user transmits single stream data, V_pIs composed ofMaximum eigenvalue λ of_maxCorresponding feature vector x, when R_picoObtaining a maximum value R_picomax，R_picomax＝log(P_pλ_max+1)；

If the micro-user transmits n stream data, selectingMaximum n eigenvalues λ₁,λ₂,…λ_nCorresponding orthogonal eigenvectors x₁,x₂,…x_nObtaining a precoding matrix V as a beamforming vector for each stream data_p＝[x₁x₂…x_n]At this time, R_picoObtaining a maximum value R_picomax， $R_{picomax}=\underset{k=1}{\overset{n}{\Σ}}log(P_p{\mathrm{\λ}}_k+1).$

5. The method according to claim 1, wherein in step 2.2, when the micro user performs active interference cancellation to suppress interference of the macro base station, the sum data rate R is:

R＝R_macro+R_pico,

R_macrodata rate, R, of macro base station_picoRepresenting the data rate of the micro-user;

in order to ensure that the data of the macro base station can be solved in the direct transmission link and the cross link, the data rate of the macro base station meets the following conditions:

$\left\{\begin{array}{c}{R}_{macro}\≤\log |I+\frac{P_m{H}_{mm}{V}_m{V}_m^H{H}_{mm}^H}{I+P_p{H}_{mp}{V}_p{V}_p^H{H}_{mp}^H}|,\\ R_{macro}\≤\log |I+\frac{P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H}{I+P_p{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H}|,\end{array}\right.$

thus, there are

$R_{macro}=\log \min \left(\right|I+\frac{P_m{H}_{mm}{V}_m{V}_m^H{H}_{mm}^H}{I+P_p{H}_{mp}{V}_p{V}_p^H{H}_{mp}^H}|,|I+\frac{P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H}{I+P_p{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H}\left|\right)$

Because the micro base station is closer to the micro user than the macro base station, when the cross link interference is solved, the signal transmitted to the micro user by the micro base station is taken as the interference, and at the moment, the data of the micro base station forms strong interference; assuming that the channel environment of the cross-link is worse, the macro-user data rate is further expressed as

$R_{macro}=\log |I+\frac{P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H}{I+P_p{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H}|,$

And due to

$R_{pico}=log|I+P_p{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H|,$

Thus, the resulting sum data rate R is:

$R=R_{macro}+R_{pico}=log|I+P_p{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H|,$

further transformation of the above equation yields:

$\begin{array}{c}R=log\left|(I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H)(I+P_p{\left(I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H\right)}^{-1}{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H)\right|\\ =log|I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H||I+P_p{\left(I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H\right)}^{-1}{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H)|\\ =log|I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H|+\log |I+P_p{\left(I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H\right)}^{-1}{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H)|\end{array},$

will be regular matrixIs decomposed into

Then the sum data rate R is further expressed as:

$\begin{array}{c}R=log|I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H|+\log |I+P_p{Q}_2^H{Q}_2{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H)|\\ =log|I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H|+\log |I+P_p{Q}_2{H}_{pp}{V}_p{V}_p^H{H}_{pp}^H{Q}_2^H)|\end{array},$

when a micro-user transmits single-stream data, set V_pIs composed ofMaximum eigenvalue λ of_maxCorresponding feature vector x, when R takes the maximum value R_max， $R_{max}=log|I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H|+log(P_p{\mathrm{\λ}}_{max}+1);$

When the micro-user transmits n stream data, n is an integer greater than 1, selectingOrthogonal eigenvector x corresponding to the largest n eigenvalues₁,x₂,…x_nForming a precoding matrix V_pWhen R takes the maximum value R_max：

$R_{max}=log|I+P_m{H}_{pm}{V}_m{V}_m^H{H}_{pm}^H|+\underset{k=1}{\overset{n}{\Σ}}log(P_p{\mathrm{\λ}}_k+1)$

Wherein λ is₁,λ₂,…λ_nArranged in descending order, x₁,x₂,…x_nAre each lambda₁,λ₂,…λ_nThe corresponding feature vector.