Background technology
MIMO (Multi-input Multi-output, multiple-input and multiple-output) technology is one of study hotspot of wireless communication field in recent years, it can utilize multiple antennas of transmitting terminal independently to send signal separately, utilizes multiple antenna receive and recover prime information at receiving terminal simultaneously.Extensive antenna (Massive MIMO) adopts the MIMO technology of large-scale antenna array at receiving terminal, can make full use of spatial degrees of freedom and improve space division multiplexing gain.
The realization of space division multiplexing gain depends on receiving terminal and knows downlink channel condition information (ChannelState Information, CSI).At time division multiplexing (Time Division Duplexing, TDD) in system, the transmitting and receiving of signal are carried out in same frequency channel, and receiving terminal can utilize channel reciprocity by estimating uplink channel status information, are similar to and obtain downlink channel condition information.And in Frequency Division Duplexing (FDD) (Frequency Divisi on Duplexing, the FDD) system that range of application is wider, channel reciprocity is no longer set up, receiving terminal needs by up channel feeding back channel state information, i.e. Limited Feedback mechanism.
In traditional MIMO Limited Feedback mechanism, first receiving terminal utilizes code book to quantize channel condition information, and this code book is that sending and receiving both sides are common, then feeds back the code word sequence number in the codebook of most match channels state information.For realizing the gain of mimo system space division multiplexing completely, the size of channel quantitative code book exponentially increases with number of transmit antennas.Codebook search complexity when this had both added receiving terminal quantized channel, has increased the weight of again the feedback burden of up link.In extensive antenna (Massive MIMO) system, antenna amount is huge, and traditional MIMO Limited Feedback mechanism will inevitably increase codebook search complexity and feedback burden.
At present, compressed sensing (Compressive Sensing, CS) technology is utilized can to compress MassiveMIMO channel condition information.In large-scale antenna array, closely, spacing is generally half-wavelength, has spatial coherence, and compressed sensing make use of this spatial coherence just in each bay arrangement.Compressed sensing technology is pointed out: for the channel matrix with spatial coherence, can find one group of orthogonal basis Ψ, makes channel matrix be sparse on Ψ.This is the basis that compressed sensing is achieved.Then receiving terminal is measured channel matrix with the incoherent observing matrix Φ of Ψ with one, high-dimensional channel condition information is projected in lower dimensional space, transmitting terminal, by solving an optimization problem, can restore high-dimensional channel condition information from these a small amount of projections high probability.
But, after utilizing compressed sensing to compress higher-dimension channel information, be the mode by analog feedback, namely send the channel condition information after compression directly to receiving terminal on the uplink channel.Be different from the Limited Feedback mode based on quantizing code book, analog feedback requires that first receiving terminal sends uplink detection reference signal (Sounding Reference Signal, SRS), receiving terminal also will utilize SRS to estimate uplink channel status, and this brings added burden to receiving terminal and receiving terminal.Secondly, receiving terminal, after receiving the low-dimensional channel condition information fed back, also needing just can restore higher-dimension channel information by solving an optimization problem, adding process complexity.In addition, in existing compressed sensing technology, although can reduce uplink feedback burden to a certain extent, compression ratio can not be too low.For ensureing that mean square error is in the acceptable scope of system, the compression ratio of compressed sensing generally should 10%---in 30% scope.
Summary of the invention
The object of the application is, provide a kind of limited feedback method and device of extensive antenna system, to solve the problem of above-mentioned technology, concrete scheme comprises:
On the one hand, embodiments provide the limited feedback method of extensive antenna system, comprising:
Described receiving terminal feeds back the low-dimensional channel condition information utilizing Principal Component Analysis Method to obtain in each short period to described transmitting terminal;
Described receiving terminal, in each long period, continues to carry out channel estimating, and utilizes Principal Component Analysis Method to obtain dimensionality reduction matrix, and to described transmitting terminal feedback dimensionality reduction matrix;
Dimensionality reduction transpose of a matrix, at each macrocyclic dimensionality reduction matrix starting to receive receiving terminal feedback, is multiplied with low-dimensional channel information, restores dimensional state information by described transmitting terminal;
Described transmitting terminal each short period receives the low-dimensional channel condition information that described receiving terminal feeds back.
Preferably, described receiving terminal feeds back to described transmitting terminal the low-dimensional channel condition information utilizing Principal Component Analysis Method to obtain and comprises in each short period:
Short period operates in each long period and repeats L time, and concrete steps are as follows:
Channel estimating is carried out to down channel, obtains higher-dimension channel condition information
i=1,2...L;
The dimensionality reduction matrix utilizing the previous long period of receiving terminal to obtain does dimension-reduction treatment to higher-dimension channel condition information h (i), obtains low-dimensional channel condition information
Feed back corresponding code word sequence number to transmitting terminal.
Preferably, described receiving terminal, in each long period, continues to carry out channel estimating, and utilizes Principal Component Analysis Method to obtain dimensionality reduction matrix, and feeds back described dimensionality reduction matrix to described transmitting terminal and specifically comprise:
Continue to carry out channel estimating in long period, obtain a large amount of higher-dimension channel condition information sample;
Principal Component Analysis Method is used to a large amount of higher-dimension channel condition information samples obtained, obtains dimensionality reduction matrix
do dimension-reduction treatment for receiving terminal at next long period, and transmitting terminal uses at next long period reduction higher-dimension channel condition information;
Based on codebook quantification dimensionality reduction matrix;
Quantitative information is fed back to transmitting terminal.
Preferably, described transmitting terminal, at each macrocyclic dimensionality reduction matrix starting to receive receiving terminal feedback, specifically comprises:
The dimensionality reduction matrix fed back from receiving terminal (receiving terminal) is received each macrocyclic beginning
Pseudoinverse being asked to U ', obtaining for reducing the recovery matrix of higher-dimension channel condition information:
Use at current long period.
Preferably, dimensionality reduction transpose of a matrix, at each macrocyclic dimensionality reduction matrix starting to receive receiving terminal feedback, is multiplied with low-dimensional channel information, restores dimensional state information, specifically comprise by described transmitting terminal:
The low-dimensional channel condition information from receiving terminal feedback is received in each short period
Matrix multiplication operation is utilized to incite somebody to action
be reduced into higher-dimension channel condition information:
Preferably, Principal Component Analysis Method mainly comprises:
Continue observation L short period (each short period s TTI), obtain L*s channel vector sample h
11h
1s, h
21h
2sh
l1h
ls;
Order
and ask its covariance matrix:
Feature decomposition is carried out to covariance matrix C: C=UDU
h;
Wherein D is diagonal matrix characteristic value being arranged from big to small formation, and U is the unitary matrice that corresponding characteristic vector is formed;
Choose M row before in U and form dimensionality reduction matrix
column vector in U ' is main characteristic vector;
Utilize dimensionality reduction matrix to carry out dimension-reduction treatment to higher-dimension channel condition information: h '=hU ', h is projected in main characteristic vector, obtains
On the other hand, embodiments provide a kind of Limited Feedback device of extensive antenna system, based on receiving terminal and transmitting terminal, described device comprises:
Low-dimensional channel condition information obtains module, is arranged in described receiving terminal, for feeding back the low-dimensional channel condition information utilizing Principal Component Analysis Method to obtain in each short period to described transmitting terminal;
Feedback dimensionality reduction matrix module, is arranged in described receiving terminal, in each long period, continues to carry out channel estimating, and utilizes Principal Component Analysis Method to obtain dimensionality reduction matrix, and to described transmitting terminal feedback dimensionality reduction matrix;
Dimensionality reduction matrix reception module, is arranged in described transmitting terminal, at each macrocyclic dimensionality reduction matrix starting to receive receiving terminal feedback;
Dimensional state information reverting module, for dimensionality reduction transpose of a matrix being multiplied with low-dimensional channel information, restores dimensional state information; Low-dimensional channel condition information receiver module, is arranged in described transmitting terminal, receives the low-dimensional channel condition information of described receiving terminal feedback for each short period.
Compared with prior art, the application comprises following advantage: utilize Principal Component Analysis Method to carry out in the process of dimension-reduction treatment to the higher-dimension channel condition information with spatial coherence, the generally concentrated expression of available 2 to 3 main compositions of higher-dimension channel condition information, reduces compression ratio.Relevant simulation result shows, the contribution rate of main composition can reach more than 95%.In addition, be the Limited Feedback mode based on low-dimensional metrization code book to the feedback of channel condition information after dimensionality reduction, receiving terminal only need feed back corresponding code word sequence number, without the need to sending SRS detectable signal, reduces receiving terminal burden.Transmitting terminal, after receiving low-dimensional channel condition information, only needs to carry out matrix multiplication operation and can restore higher-dimension channel condition information, without the need to solving complicated optimization problem, reduces the process complexity of transmitting terminal.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present application, be clearly and completely described the technical scheme in the embodiment of the present application, obviously, described embodiment is only some embodiments of the present application, instead of whole embodiments.Based on the embodiment in the application, those of ordinary skill in the art are not making the every other embodiment obtained under creative work prerequisite, all belong to the scope of the application's protection.
The embodiment of the present invention based on extensive antenna system, receiving terminal (UE) by Limited Feedback mode to transmitting terminal (base station) feedback of down channel state information.The scene Shi Dan community supposed, multi-user, base station deployment large-scale antenna array, number of transmit antennas is N
t, and user is equipped with single antenna.For certain user, carrying out at moment t the channel matrix that channel estimating obtains is a N
tdimension row vector, namely
The embodiment of the present invention proposes a kind of limited feedback method of extensive antenna system.In the present invention, different feedback cycles is used to dissimilar information.To in the embodiment of the present invention, the setting of long and short period is elaborated below.
Long period: in the present invention, UE utilizes Principal Component Analysis Method the higher-dimension channel information with spatial coherence can be represented with a few main composition is concentrated, and namely dimension-reduction treatment becomes low-dimensional channel information, and by uplink feedback to base station.In order to base station can restore higher-dimension channel condition information, UE also needs to fed back through the dimensionality reduction matrix that Principal Component Analysis Method obtains.Be different from the Real-time Feedback of low-dimensional channel information, dimensionality reduction matrix, need not Real-time Feedback owing to being carrying out channel status the basis continuing to estimate obtains a kind of statistical information, suitably can elongate feedback cycle.The present invention uses long period feedback to the dimensionality reduction matrix obtained by Principal Component Analysis Method.
Short period: in the present invention, short period feedback is used to the low-dimensional channel condition information utilizing Principal Component Analysis Method to obtain, this low-dimensional channel condition information is the concentrated expression of higher-dimension channel condition information, UE needs Real-time Feedback to base station, is convenient to base station and understands channel status in time and make respective handling.
As shown in Figure 1, time resource is divided into the elementary cell in units of the short period, and each short period comprises some TTI (Transmit Time Interval), and especially, each TTI is exactly a short period.In each short period, UE is to the low-dimensional channel condition information that base station feedback utilizes Principal Component Analysis Method to obtain.Several short periods form a long period, and in each long period, UE continues to carry out channel estimating, and utilizes Principal Component Analysis Method to obtain dimensionality reduction matrix, and the feedback of dimensionality reduction matrix is carried out with long period.So in each last short period macrocyclic, UE not only will feed back low-dimensional channel condition information, also will feed back dimensionality reduction matrix.
In addition, as can be seen from Figure 1, UE carries out the dimensionality reduction matrix used by dimension-reduction treatment at current long period to higher-dimension channel condition information, and the recovery matrix that base station is used at current long period reduction higher-dimension channel condition information, all utilize Principal Component Analysis Method to obtain by UE at previous long period.
Embodiment one
The embodiment of the present invention one provides a kind of limited feedback method of extensive antenna system, and based on extensive antenna system, extensive antenna system comprises receiving terminal and transmitting terminal, and as shown in Figure 2, method comprises:
Step S201, receiving terminal feeds back the low-dimensional channel condition information utilizing Principal Component Analysis Method to obtain in each short period to transmitting terminal;
Step S202, receiving terminal, in each long period, continues to carry out channel estimating, and utilizes Principal Component Analysis Method to obtain dimensionality reduction matrix, and to transmitting terminal feedback dimensionality reduction matrix;
Step S203, dimensionality reduction transpose of a matrix, at each macrocyclic dimensionality reduction matrix starting to receive receiving terminal feedback, is multiplied with low-dimensional channel information, restores dimensional state information by transmitting terminal;
Step S204, transmitting terminal each short period receives the low-dimensional channel condition information that receiving terminal feeds back.
Concrete, in step s 201, receiving terminal feeds back to transmitting terminal the low-dimensional channel condition information utilizing Principal Component Analysis Method to obtain and comprises in each short period:
Short period operates in each long period and repeats L time, and concrete steps are as follows:
Channel estimating is carried out to down channel, obtains higher-dimension channel condition information
i=1,2...L;
The dimensionality reduction matrix utilizing the previous long period of receiving terminal to obtain does dimension-reduction treatment to higher-dimension channel condition information h (i), obtains low-dimensional channel condition information
Based on codebook quantification low-dimensional channel condition information h (i) ', the size of code book depends on the dimension of h (i) ', relative to prior art, can codebook size be reduced, reduce receiving terminal (UE) codebook search complexity and feedback quantity.Meanwhile, the burden sending uplink detection reference signal (SRS) is eliminated;
Feed back corresponding code word sequence number to transmitting terminal.
In step S202, receiving terminal, in each long period, continues to carry out channel estimating, and utilizes Principal Component Analysis Method to obtain dimensionality reduction matrix, and specifically comprises to transmitting terminal feedback dimensionality reduction matrix:
Continue to carry out channel estimating in long period, obtain a large amount of higher-dimension channel condition information sample;
Principal Component Analysis Method is used to a large amount of higher-dimension channel condition information samples obtained, obtains dimensionality reduction matrix
do dimension-reduction treatment for receiving terminal at next long period, and transmitting terminal uses at next long period reduction higher-dimension channel condition information;
Based on codebook quantification dimensionality reduction matrix;
Quantitative information is fed back to transmitting terminal.
In step S203, transmitting terminal, at each macrocyclic dimensionality reduction matrix starting to receive receiving terminal feedback, specifically comprises:
The dimensionality reduction matrix from receiving terminal feedback is received each macrocyclic beginning
Pseudoinverse being asked to U ', obtaining for reducing the recovery matrix of higher-dimension channel condition information:
Use at current long period.
In step S204, dimensionality reduction transpose of a matrix, at each macrocyclic dimensionality reduction matrix starting to receive receiving terminal feedback, is multiplied with low-dimensional channel information, restores dimensional state information, specifically comprise by transmitting terminal:
The low-dimensional channel condition information from receiving terminal feedback is received in each short period
Matrix multiplication operation is utilized to incite somebody to action
be reduced into higher-dimension channel condition information:
Can find out, only need carry out matrix multiplication can restore higher-dimension channel condition information to transmitting terminal (base station), do not need, as solving a complicated optimization problem in compressed sensing technology, to decrease the process complexity of transmitting terminal (base station).
In embodiments of the present invention, Principal Component Analysis Method mainly comprises:
Continue observation L short period (each short period s TTI), obtain L*s channel vector sample h
11h
1s, h
21h
2sh
l1h
ls;
Order
and ask its covariance matrix:
Feature decomposition is carried out to covariance matrix C: C=UDU
h;
Wherein D is diagonal matrix characteristic value being arranged from big to small formation, and U is the unitary matrice that corresponding characteristic vector is formed;
Choose M row before in U and form dimensionality reduction matrix
column vector in U ' is main characteristic vector;
Utilize dimensionality reduction matrix to carry out dimension-reduction treatment to higher-dimension channel condition information: h '=hU ', h is projected in main characteristic vector, obtains
The limited feedback method of a kind of extensive antenna system provided by the embodiment of the present invention, Principal Component Analysis Method is utilized to carry out in the process of dimension-reduction treatment to the higher-dimension channel condition information with spatial coherence, the generally concentrated expression of available 2 to 3 main compositions of higher-dimension channel condition information, reduces compression ratio.Relevant simulation result shows, the contribution rate of main composition can reach more than 95%.In addition, be the Limited Feedback mode based on low-dimensional metrization code book to the feedback of channel condition information after dimensionality reduction, receiving terminal only need feed back corresponding code word sequence number, without the need to sending SRS detectable signal, reduces receiving terminal burden.Transmitting terminal, after receiving low-dimensional channel condition information, only needs to carry out matrix multiplication operation and can restore higher-dimension channel condition information, without the need to solving complicated optimization problem, reduces the process complexity of transmitting terminal.
Embodiment two
The embodiment of the present invention two provides a kind of Limited Feedback device of extensive antenna system, based on receiving terminal and transmitting terminal, as shown in Figure 3, described device comprises: low-dimensional channel condition information obtains module 31, feedback dimensionality reduction matrix module 32, dimensionality reduction matrix reception module 33, dimensional state information reverting module 34 and low-dimensional channel condition information receiver module 35.Concrete:
Low-dimensional channel condition information obtains module 31, is arranged in described receiving terminal, for feeding back the low-dimensional channel condition information utilizing Principal Component Analysis Method to obtain in each short period to described transmitting terminal;
Feedback dimensionality reduction matrix module 32, is arranged in described receiving terminal, in each long period, continues to carry out channel estimating, and utilizes Principal Component Analysis Method to obtain dimensionality reduction matrix, and to described transmitting terminal feedback dimensionality reduction matrix;
Dimensionality reduction matrix reception module 33, is arranged in described transmitting terminal, at each macrocyclic dimensionality reduction matrix starting to receive receiving terminal feedback;
Dimensional state information reverting module 34, for dimensionality reduction transpose of a matrix is multiplied with low-dimensional channel information, restore dimensional state information low-dimensional channel condition information receiver module 35, be arranged in described transmitting terminal, receive the low-dimensional channel condition information of described receiving terminal feedback for each short period.
A kind of Limited Feedback device of extensive antenna system is provided by the embodiment of the present invention two, Principal Component Analysis Method is utilized to carry out in the process of dimension-reduction treatment to the higher-dimension channel condition information with spatial coherence, the generally concentrated expression of available 2 to 3 main compositions of higher-dimension channel condition information, reduces compression ratio.Relevant simulation result shows, the contribution rate of main composition can reach more than 95%.In addition, be the Limited Feedback mode based on low-dimensional metrization code book to the feedback of channel condition information after dimensionality reduction, receiving terminal only need feed back corresponding code word sequence number, without the need to sending SRS detectable signal, reduces receiving terminal burden.Transmitting terminal, after receiving low-dimensional channel condition information, only needs to carry out matrix multiplication operation and can restore higher-dimension channel condition information, without the need to solving complicated optimization problem, reduces the process complexity of transmitting terminal.
Above image characteristic extracting method, server, terminal and system that the application provides are described in detail, apply specific case herein to set forth the principle of the application and execution mode, the explanation of above embodiment is just for helping method and the core concept thereof of understanding the application; Meanwhile, for one of ordinary skill in the art, according to the thought of the application, all will change in specific embodiments and applications, to sum up, this description should not be construed as the restriction to the application.