CN104683074B - Extensive mimo system limited feedback method based on compressed sensing - Google Patents
Extensive mimo system limited feedback method based on compressed sensing Download PDFInfo
- Publication number
- CN104683074B CN104683074B CN201510111378.3A CN201510111378A CN104683074B CN 104683074 B CN104683074 B CN 104683074B CN 201510111378 A CN201510111378 A CN 201510111378A CN 104683074 B CN104683074 B CN 104683074B
- Authority
- CN
- China
- Prior art keywords
- sparse
- channel
- indicate
- extensive mimo
- signal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Mobile Radio Communication Systems (AREA)
- Radio Transmission System (AREA)
Abstract
A kind of extensive mimo system limited feedback method based on compressed sensing, in extensive MIMO Limited feedback systems, base station end uses Nh×NvUniform surface battle array (Uniform Planar Array, UPA) is tieed up, user terminal uses linear array.First, utilize compressed sensing (Compressive Sensing, CS) theoretical to find the sparse basis to match with vertical dimension channel with extensive MIMO levels dimension channel, and then it is accumulated to obtain the two-dimentional joint sparse base to match with extensive mimo channel according to Kronecker, then, condition of sparse channel is projected to lower dimensional space using with sparse basis incoherent observing matrix, to obtain observation signal, and by this signal it is quantified after feed back to base station, finally, pass through orthogonal tracking matching algorithm (Orthogonal Matching Pursuit, OMP) channel state information is rebuild in base station end.The invention provides a kind of simple and efficient implementation method for extensive MIMO Limited feedback systems.
Description
Technical field
The present invention relates to 5G mobile communication technology fields, show in particular a kind of extensive MIMO systems based on compressed sensing
System limited feedback method.
Background technology
With the sharp increase of amount of communication data, current Development of Wireless Communications is become to the raising of the availability of frequency spectrum and efficiency
Key.It is well known that MIMO (Massive-Multiple Input Multiple Output, Massive MIMO, it is big to advise
Mould multiple-input and multiple-output) technology can improve link transmission reliability and obtain higher power system capacity, therefore it is wireless to become 4G
The key technology of communication system.The capacity of mimo system increases with the increase of number of antennas, therefore extensive MIMO is in base
End of standing is configured with the antenna of huge number, to obtain better spatial reuse and space diversity effect, and can be more effective
Eliminate user between interference.From the perspective of energy efficiency, the wave beam that extensive MIMO is formed is relatively narrow, can make transmitting work(
Rate focuses more on specific user, so as to obtain higher receiving terminal signal-to-noise ratio using smaller transmission power.Because big
Scale MIMO has the advantages that low energy consumption and high performance, so receiving the extensive concern of industry and the crucial research skill as 5G
Art.Since the extensive MIMO antenna numbers configured in base station end increase, simultaneously to traditional limited feedback method based on code book
Extensive mimo system is not suitable for it, it is therefore desirable to design effective feedback algorithm and be opened to reduce the feedback of extensive mimo system
Pin.
There is a kind of new theory-compressed sensing in signal processing and the communications field in recent years, to study the pressure of channel
Contracting feedback provides new thinking, which points out:As long as signal is compressible or is sparse in some transform domain, then
The high dimensional signal of gained can will be converted with one with the transformation incoherent observing matrix of base to project on a lower dimensional space, so
Original signal can be reconstructed with high probability from these a small amount of projections by solving an optimization problem afterwards, can prove this
The projection of sample contains the enough information of reconstruction signal.
The rarefaction representation of signal is basis and the premise of compressive sensing theory application, has only reasonably selected sparse basis,
It can just make the sparse coefficient number of signal as few as possible, to ensure signaling protein14-3-3 precision.Common Ka Luonan-Luo Yi
(Karhunen-Loeve Transform, KLT) transformation is a transformation with lowest mean square meaning under statistical significance, tool
There are optimal decorrelation and extraordinary compression efficiency, but be a lack of fast algorithm, and transformation matrix is different with signal, it is different
Signal need to calculate different transformation matrixs, therefore computation complexity is larger.FFT has efficient algorithm, so FFT is in addition to resonable
By upper particularly significant outer, central role is also acted as in realizing various digital signal processing algorithms.DCT has very strong " energy
Concentrate " characteristic:Most of sound and the energy of picture signal all concentrate on the low frequency part after discrete cosine transform, and work as
When signal is had close to the statistical property of markoff process, the decorrelation of DCT is close to Karhunen-Loeve transformation.In extensive MIMO
In system, the Limited Feedback technology based on compressed sensing, academic at present is in the Primary Study stage with industrial circle.Due to big
Scale MIMO base station antenna number is excessive, and uses UPA, therefore the Limited Feedback technical research based on compressed sensing not only needs to examine
How worry finds suitable sparse basis, it is also necessary to effectively reduce feedback overhead.
Invention content
For the above the deficiencies in the prior art, the purpose of the present invention is to provide a kind of based on the extensive of compressed sensing
Mimo system limited feedback method, this method can not only find the two-dimentional joint sparse base for being suitable for extensive mimo system, also
Channel information can be recovered in base station end, while reduce feedback overhead.
Technical scheme is as follows:
A kind of extensive mimo system limited feedback method based on compressed sensing comprising following steps:
Step 101, base station end are configured with Nh×NvThe UPA of dimension, wherein NhFor the horizontal dimension transmitting antenna number of UPA, NvFor
The vertical dimension transmitting antenna number of UPA;User terminal is configured to single antenna, and channel is represented byWherein hhIt indicates
Level dimension channel information, hvIndicate vertical dimension channel information.
The horizontal dimension sparse basis Ψ of step 102, design HhWith vertical dimension sparse basis Ψv, wherein ΨhBecome using discrete cosine
Change (discrete cosine transform, DCT) sparse basis, ΨvUsing Fast Fourier Transform (FFT) (Fast Fourier
Transform, FFT) sparse basis, obtain two-dimentional joint sparse base using Kronecker product theorems
Step 103 generates random Gaussian observing matrix Φ ∈ RL×N, wherein element all obeys 0 mean value, and variance is
Independent normal distribution, wherein N=Nh× Nv, L indicate observation, and L < N.
Channel H is projected to lower dimensional space by step 104 by observing matrix Φ, obtains the observation signal y of L × N-dimensional.
Step 105, base station end are according to observation signal y, observing matrix Φ and two-dimentional joint sparse base Ψ, using OMP algorithms
Correctly recover channel information
Further, horizontal dimension channel rarefaction representation is as follows in step 102:.hh=Ψhh1., wherein h1It indicates by orthogonal
Signal after sparse basis transformation, DCT sparse basises ΨhSpecific formula it is as follows:
Contravariant
Changing IDCT formula is:
Wherein
NhFor the horizontal dimension transmitting antenna number of UPA, i, j are space coordinates, i, j=0,1 ... Nh-1;U, v are DCT space coordinates, u, v
=0,1 ... Nh-1.Variable coefficient
Similarly, vertical dimension channel rarefaction representation is as follows:hv=Ψvh2, wherein h2It indicates after the transformation of orthogonal sparse basis
Signal, FFT sparse basises ΨvSpecific formula it is as follows:
Wherein NvFor the vertical dimension transmitting antenna number of base station end uniform surface battle array,
And
Theorem is accumulated according to Kronecker:And it level dimension channel and hangs down
The rarefaction representation of straight dimension channel substitutes into extensive mimo channel model, and then obtains two-dimentional joint sparse base Ψ, specific to indicate
It is as follows:
Wherein H indicates the extensive mimo channel information of N × 1, hhIndicate Nh× 1 extensive MIMO levels dimension channel letter
Breath, hvIndicate Nv× 1 extensive MIMO vertical dimensions channel information, ΨhIndicate Nh×NhDCT sparse basises, ΨhIndicate Nh×Nh
FFT sparse basises, Ψ indicate N × N two dimension joint sparse bases, h1,h2, signal of the h expressions after the transformation of orthogonal sparse basis.
Further, there are one important properties for the random gaussian matrix Φ tools in step 103:For L × N with
Machine Gaussian matrix Φ, as L >=cKlog (N/K), Φ Ψ have RIP properties under very maximum probability:
Wherein δ ∈ (0,1), c indicate that the constant of a very little, K indicate that the degree of rarefication of channel, N indicate the length of channel, and
N=Nh×Nv。
Further, observation signal can be expressed as form in step 104:
Y=Φ H=Φ Ψ h
Wherein Φ indicates that the random gaussian matrix of L × N, H indicate that the extensive mimo channel information of N × 1, Ψ indicate N × N
Two-dimentional joint sparse base, h indicate the signal after the transformation of two-dimentional joint sparse base.
Further, the channel recovered in step 105Normalization mean square error between extensive mimo channel H
Poor (Mean Squared Error, MSE), can be acquired by following formula:
The present invention considers the excessive feature of extensive mimo antenna number, proposes the Limited Feedback side based on compressed sensing
Method finds the two-dimentional joint sparse base to match with extensive mimo channel using CS theories first, then, utilizes observation square
Condition of sparse channel is carried out linear projection by battle array from higher-dimension to low-dimensional, finally, in base station end, restores the channel of feedback by OMP algorithms
Status information.Extensive mimo system limited feedback method proposed by the present invention based on compressed sensing, compared to traditional
MIMO Limited Feedback technologies not only more match extensive mimo channel, but also can effectively reduce feedback overhead and meter
Complexity is calculated, thus, which provides a kind of simple and efficient implementation for extensive MIMO Limited Feedbacks.
Description of the drawings
Fig. 1 is the system block diagram of the extensive mimo system limited feedback method proposed by the present invention based on compressed sensing;
Fig. 2 is the uniform surface battle array schematic diagram that base station uses vertical polarized antenna;
Fig. 3 is compressed sensing basic operation flow chart;
Fig. 4 is that the present invention is based on the Limited Feedback flow charts of compressed sensing.
Specific implementation mode
In extensive MIMO Limited Feedbacks scheme, condition of sparse channel is projected to by observing matrix by lower dimensional space, and handle
Obtained observation signal feeds back to base station, and base station rebuilds the channel state information of feedback using OMP algorithms, in this way, can reduce
Computation complexity and feedback overhead.
The system for the extensive mimo system Limited Feedback based on compressed sensing that as shown in FIG. 1, FIG. 1 is proposed by the present invention
Block diagram.N is configured in base station endh×NvThe UPA of dimension, wherein NhFor the horizontal dimension transmitting antenna number of UPA, NvIt is sent out for the vertical dimension of UPA
Antenna number is penetrated, receiving terminal configures M user of single antenna, and user terminal carries out channel estimation and obtains 3D channel matrix Hs, and H can
It is expressed asWherein hhIndicate horizontal dimension channel information, hvIndicate vertical dimension channel information.
Fig. 2 is uniform surface battle array model proposed by the present invention, it is contemplated that the limitation of antenna physical size, the model as shown in the figure
Using UPA structures.
Fig. 3 is compressed sensing basic operation flow chart.As shown, compressive sensing theory includes mainly three parts, first,
Then the rarefaction representation of signal designs observing matrix, to ensure the information loss of original signal most while reducing dimension
Small, finally, modelled signal recovery algorithms recover original signal without distortions using observation.
Fig. 4 is that the present invention proposes the Limited Feedback flow chart based on compressed sensing.
Step 41:Base station end configures Nh×NvTie up the UPA, wherein N of vertical polarized antennahFor the horizontal dimension transmitting antenna of UPA
Number, NvFor the vertical dimension transmitting antenna number of UPA, receiving terminal configures M user of single antenna.
Step 42:Dct transform is carried out to level dimension channel, i.e.,:
hh=Ψhh1
Wherein h1Indicate the signal after the transformation of orthogonal sparse basis, DCT sparse basises ΨhSpecific formula it is as follows:
Contravariant
Changing IDCT formula is:
Wherein NhFor the horizontal dimension transmitting antenna number of uniform surface battle array UPA, i, j are space coordinates, i, j=0,1 ... Nh-1;u,v
It is DCT space coordinates, u, v=0,1 ... Nh-1.Variable coefficient
Step 43:Orthogonal sparse basis transformation is carried out to vertical dimension channel, obtains sparse signal h2, i.e.,:
hv=Ψvh2
Wherein FFT sparse basises ΨvSpecific formula it is as follows:
Wherein
And
Step 44:Theorem is accumulated according to Kronecker:And level is tieed up
The rarefaction representation of channel and vertical dimension channel substitutes into extensive mimo channel model, and then obtains two-dimentional joint sparse base Ψ,
It is specific to indicate as follows:
Wherein H indicates the extensive mimo channel information of N × 1, hhIndicate Nh× 1 extensive MIMO levels dimension channel letter
Breath, hvIndicate Nv× 1 extensive MIMO vertical dimensions channel information, ΨhIndicate Nh×NhDCT sparse basises, ΨhIndicate Nh×Nh
FFT sparse basises, Ψ indicate N × N two dimension joint sparse bases, h1,h2, signal of the h expressions after the transformation of orthogonal sparse basis.
Step 45:Random gaussian matrix Φ is generated, and must be had there are one important properties:For the random of L × N
Gaussian matrix Φ, for the random gaussian matrix Φ of a L × N, as L >=cKlog (N/K), Φ Ψ have under very maximum probability
RIP properties:
Wherein δ ∈ (0,1), c indicate that the constant of a very little, K indicate that the degree of rarefication of channel, N indicate the length of channel, and
N=Nh×Nv, signal of the h expressions after the transformation of two-dimentional joint sparse base.
Step 46:Using the observing matrix Φ of L × N-dimensional channel is subjected to linear projection from higher-dimension to low-dimensional, obtains L × N
The observation signal y of dimension:
Y=Φ H=Φ Ψ h
Wherein Φ indicates that the random gaussian matrix of L × N, H indicate that the extensive mimo channel information of N × 1, Ψ indicate N × N
Two-dimentional joint sparse base, h indicate the signal after the transformation of two-dimentional joint sparse base.
Step 47:Base station end is extensive using OMP algorithms according to observation signal y, observing matrix Φ and two-dimentional joint sparse base Ψ
It appears again channel informationAnd the normalization MSE asked by following formula:
As it can be seen that the present invention provides a kind of and simple and efficient implementation method for extensive MIMO Limited feedback systems.
Claims (5)
1. a kind of extensive mimo system limited feedback method based on compressed sensing, it is characterised in that:Include the following steps:
Step 101, base station end are configured with Nh×NvThe UPA of dimension, wherein NhFor the horizontal dimension transmitting antenna number of UPA, NvFor UPA's
Vertical dimension transmitting antenna number;User terminal is configured to single antenna, and channel is expressed asWherein hhIndicate horizontal dimension letter
Road information, hvIndicate vertical dimension channel information;
The horizontal dimension sparse basis Ψ of step 102, design HhWith vertical dimension sparse basis Ψv, wherein ΨhUsing discrete cosine transform
Sparse basis, ΨvUsing Fast Fourier Transform (FFT) FFT sparse basises, two-dimentional joint sparse base is obtained using Kronecker product theorems
Step 103 generates random Gaussian observing matrix Φ ∈ RL×N, wherein element all obeys 0 mean value, and variance isIt is only
Vertical normal distribution, wherein N=Nh×Nv, L expression observations, and L < N;
Channel H is projected to lower dimensional space by step 104 by observing matrix Φ, obtains the observation signal y of L × N-dimensional;
Step 105, base station end are restored according to observation signal y, observing matrix Φ and two-dimentional joint sparse base Ψ using OMP algorithms
Go out channel information
Horizontal dimension channel rarefaction representation is as follows in step 102:hh=Ψhh1, wherein h1It indicates after the transformation of orthogonal sparse basis
Signal, ΨhIndicate DCT sparse basises;
Vertical dimension channel rarefaction representation is as follows:hv=Ψvh2, wherein h2Indicate the signal after the transformation of orthogonal sparse basis, ΨvTable
Show FFT sparse basises.
2. according to the method described in claim 1, it is characterized in that, accumulating theorem according to Kronecker:
Horizontal peacekeeping vertical dimension channel is brought intoTo
To two-dimentional joint sparse base Ψ, i.e.,
Wherein h indicates the condition of sparse channel after the transformation of two-dimentional joint sparse base.
3., will be dilute using the observing matrix Φ of L × N-dimensional according to the method described in claim 1, it is characterized in that, in step 103
It dredges channel and carries out linear projection from higher-dimension to low-dimensional, the product of observing matrix and sparse basis array must satisfy limited equidistant item
Part RIP, to guarantee the accurate reconstruction signal from observation signal, RIP conditions are as follows:
Wherein δ ∈ (0,1), h indicate the condition of sparse channel after the transformation of two-dimentional joint sparse base.
4. according to the method described in claim 1, it is characterized in that, observation signal is expressed as form in step 104:
Y=Φ H=Φ Ψ h
Wherein Φ indicates that the random gaussian matrix of L × N, H indicate that the extensive mimo channel information of N × 1, Ψ indicate N × N two dimensions
Joint sparse base, h indicate the signal after the transformation of two-dimentional joint sparse base.
5. according to the method described in claim 1, it is characterized in that, the channel recovered in step 105With extensive MIMO
Normalized mean squared error MSE between channel H, is acquired by following formula:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510111378.3A CN104683074B (en) | 2015-03-13 | 2015-03-13 | Extensive mimo system limited feedback method based on compressed sensing |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510111378.3A CN104683074B (en) | 2015-03-13 | 2015-03-13 | Extensive mimo system limited feedback method based on compressed sensing |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104683074A CN104683074A (en) | 2015-06-03 |
CN104683074B true CN104683074B (en) | 2018-09-11 |
Family
ID=53317718
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510111378.3A Active CN104683074B (en) | 2015-03-13 | 2015-03-13 | Extensive mimo system limited feedback method based on compressed sensing |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104683074B (en) |
Families Citing this family (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106209193B (en) * | 2016-08-19 | 2019-05-24 | 西华大学 | CSI feedback method based on the extensive mimo system of compressed sensing |
CN106374936B (en) * | 2016-09-05 | 2019-09-10 | 广东顺德中山大学卡内基梅隆大学国际联合研究院 | A kind of power grid real-time control method based on compression sensing technology |
CN106506415B (en) * | 2016-11-25 | 2019-05-10 | 重庆邮电大学 | A kind of method of multi-user MIMO-OFDM system channel estimation |
EP3565133B1 (en) * | 2017-01-26 | 2021-04-21 | Huawei Technologies Co., Ltd. | Method for feeding back channel state information, terminal device, and network device |
CN107911153B (en) * | 2017-10-31 | 2021-01-19 | 东南大学 | FDD system-oriented downlink channel reconstruction method based on uplink CSI |
CN108847876B (en) * | 2018-07-26 | 2021-03-02 | 东南大学 | Large-scale MIMO time-varying channel state information compression feedback and reconstruction method |
CN109104229B (en) * | 2018-08-13 | 2022-01-11 | 南京邮电大学 | Large-scale MIMO channel feedback reconstruction algorithm based on compressed sensing |
CN109921799B (en) * | 2019-02-20 | 2023-03-31 | 重庆邮电大学 | Tensor compression method based on energy-gathering dictionary learning |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1841985A (en) * | 2005-03-30 | 2006-10-04 | 松下电器产业株式会社 | Limited feedback method for multi-antenna system |
CN103929281A (en) * | 2014-04-02 | 2014-07-16 | 同济大学 | Vector-auto-regression-based Bayes compressed sensing feedback method in MIMO system |
CN104168047A (en) * | 2014-08-13 | 2014-11-26 | 电子科技大学 | Single-ended time domain beam searching method based on compressed sensing |
-
2015
- 2015-03-13 CN CN201510111378.3A patent/CN104683074B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1841985A (en) * | 2005-03-30 | 2006-10-04 | 松下电器产业株式会社 | Limited feedback method for multi-antenna system |
CN103929281A (en) * | 2014-04-02 | 2014-07-16 | 同济大学 | Vector-auto-regression-based Bayes compressed sensing feedback method in MIMO system |
CN104168047A (en) * | 2014-08-13 | 2014-11-26 | 电子科技大学 | Single-ended time domain beam searching method based on compressed sensing |
Non-Patent Citations (2)
Title |
---|
"Compressive Sensing Based Channel Feedback Protocols for Spatially-Correlated Massive Antenna Arrays";Ping-Heng Kuo,et al;《2012 IEEE Wireless Communications and Networking Conference: PHY and Fundamentals》;20121231;全文 * |
"基于空频相关性的大规模MIMO-OFDM信道压缩反馈算法";李晓辉等;《电子与信息学报》;20140531;第36卷(第5期);第1178-1183页 * |
Also Published As
Publication number | Publication date |
---|---|
CN104683074A (en) | 2015-06-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104683074B (en) | Extensive mimo system limited feedback method based on compressed sensing | |
CN109104229B (en) | Large-scale MIMO channel feedback reconstruction algorithm based on compressed sensing | |
WO2006007148A1 (en) | Closed loop mimo systems using codebooks for feedback | |
CN104935367B (en) | A kind of iteratively faster beam-forming method based on channel space sparse characteristic | |
CN111555781B (en) | Large-scale MIMO channel state information compression and reconstruction method based on deep learning attention mechanism | |
CN109687897A (en) | Superposition CSI feedback method based on the extensive mimo system of deep learning | |
CN105162556B (en) | Extensive mimo system channel feedback method based on space-time correlation | |
CN105791189B (en) | A kind of sparse coefficient decomposition method improving reconstruction accuracy | |
CN105978655B (en) | A kind of parcel data transmission method in wireless network and system | |
CN105245263A (en) | Compressive sensing based downlink channel state information acquisition method | |
CN104779988A (en) | Quick iteration beam forming method | |
CN103546210A (en) | Pre-coding method, based on safety speed optimization, in multi-base-station cooperation scenes | |
CN103607262A (en) | Two-stage pre-coding method in space-time block coding MIMO system | |
CN103227769A (en) | Novel method for reducing peak-to-average ratio of STBC MIMO-OFDM system | |
CN105100810B (en) | Compression of images decompressing method and system in a kind of imaging sonar real time processing system | |
CN106470173A (en) | A kind of BBU-RRU data compression method | |
CN103248460B (en) | Under imperfect channel state in mimo system based on interference alignment signal processing method | |
CN105530012B (en) | The compressed sensing based sparse one-dimensional well data of wavelet field compresses and reconstructing method | |
TW201944745A (en) | Feedback method for use as a channel information based on deep learning | |
CN106788633B (en) | Uplink incoherent space-time transmission method for large-scale antenna system | |
CN107426119B (en) | Cross-shaped channel estimation method, device and system for millimeter wave communication | |
WO2016045342A1 (en) | Feedback method, apparatus and system for channel information | |
CN104540147B (en) | One kind cooperation compressed sensing quantifies retransmission method | |
CN104767557B (en) | Adaptive beam weighted information grouping feedback method based on channel quality and device | |
CN109639607B (en) | Low-complexity signal detection method for non-orthogonal multiple access system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |