Introduction to Hybrid Beamforming Techniquesaccess.ee.ntu.edu.tw/course/under_project_1032/... ·...

17
Introduction to Hybrid Beamforming Techniques Graduate Institute of Electronics Engineering National Taiwan University Taipei, Taiwan James Chen Mar. 31, 2015 Advisor : Andy Wu

Transcript of Introduction to Hybrid Beamforming Techniquesaccess.ee.ntu.edu.tw/course/under_project_1032/... ·...

Page 1: Introduction to Hybrid Beamforming Techniquesaccess.ee.ntu.edu.tw/course/under_project_1032/... · Introduction to Hybrid Beamforming Techniques Graduate Institute of Electronics

Introduction to Hybrid Beamforming Techniques

Graduate Institute of Electronics Engineering

National Taiwan University

Taipei, Taiwan

James Chen

Mar. 31, 2015

Advisor : Andy Wu

Page 2: Introduction to Hybrid Beamforming Techniquesaccess.ee.ntu.edu.tw/course/under_project_1032/... · Introduction to Hybrid Beamforming Techniques Graduate Institute of Electronics

ACCESS

Outline

Introduction of Precoding

Why Hybrid beamforming?

Problem Formulation

Existing Hybrid Beamforming Technique

Summary

2

Page 3: Introduction to Hybrid Beamforming Techniquesaccess.ee.ntu.edu.tw/course/under_project_1032/... · Introduction to Hybrid Beamforming Techniques Graduate Institute of Electronics

ACCESS

Precoding mitigates channel interference

SVD is the optimal method but require higher bandwidth

Precoding

Reduce the interference among antennas

.

.

.

.

.

.

Transmit Antennas

ReceiveAntennas

.

.

.

.

.

.

.

.

.

.

.

.

TransmitBeamforming(Precoder)

ReceiveBeamforming(Combiner)

Transmit Antennas

ReceiveAntennas

EquivalenceChannel

.

.

.

.

.

.

Transmit Antennas

ReceiveAntennas

Introduction of Precoding MIMO System

3

Vx yUVH

UH

σ1

σ4

ChannelPrecoder

SVDFeedback link

H (from RX)

Noise Decoder

RXSVD:H=UΣVH

u1 u2 u3

v1H

v2H

v3H

σ1

σ2

σ3

H =

U Σ VH

Page 4: Introduction to Hybrid Beamforming Techniquesaccess.ee.ntu.edu.tw/course/under_project_1032/... · Introduction to Hybrid Beamforming Techniques Graduate Institute of Electronics

ACCESS

Why Hybrid beamforming?(1/2)

BS

MS

In mmWave scenario, the pathloss is extremely high[3]

30 GHz shows additional about 20 dB loss compared to 3 GHz.

High pathloss can be compensated by:

Large antenna array to increase the array gain

Beamforming via precoding

Channel is rank deficient

Maximum supportable streams are less then the number of Tx antennas

4

Page 5: Introduction to Hybrid Beamforming Techniquesaccess.ee.ntu.edu.tw/course/under_project_1032/... · Introduction to Hybrid Beamforming Techniques Graduate Institute of Electronics

ACCESS

Why Hybrid beamforming?(2/2)

Traditional Beamforming is done at BB

Requiring one RF chain per transmitting antenna

A RF chain consists of a mixer, PA/LNA and DAC/ADC

Hybrid Beamforming relies on RF precoding to reduce the number of RF chains[2]

Two-staged transmitting (FRF,FBB) structure

5

Page 6: Introduction to Hybrid Beamforming Techniquesaccess.ee.ntu.edu.tw/course/under_project_1032/... · Introduction to Hybrid Beamforming Techniques Graduate Institute of Electronics

ACCESS

Problem Formulation(1/3)

Step 1: The optimal solution of the precoding matrix, Fopt ,is given by:

V1 is eigenvectors corresponding to Ns largest eigenvalues of H

V1 can be acquired from performing SVD on H

Step 2: We further realize Fopt by hybrid precoder (FRF,FBB)

Number of RF chains

can be reduced

1optF V

……

Ba

se

ba

nd

Pre

co

de

r

Ba

se

ba

nd

Eq

ua

lize

r

……

RF

Be

am

form

erRF-Chain

RF-Chain

RF-Chain

RF-Chain

……

RF-Chain

RF-Chain

RF-Chain

RF-Chain

CSI

AcquisitionSpatially Sparse Precoding

SL- SVD

Tx Precoding for Hybrid Beamformer

RFF

MIMO

Channel

H

1V

BBF

RFW

BBW

AoD

H

…… ……

……

……

……

BBF

RFF

FBBRFopt

BBRF

RFBB FFFFF

FF ,

minarg),(

6

Page 7: Introduction to Hybrid Beamforming Techniquesaccess.ee.ntu.edu.tw/course/under_project_1032/... · Introduction to Hybrid Beamforming Techniques Graduate Institute of Electronics

ACCESS

Problem Formulation(2/3)

Step 1: Get the optimal FOPT

The channel matrix H[3]:

aBS(ɵ𝑙𝐵𝑆) is the AOD of active path :

Fopt=V1 can be formed by linear combinations of aBS(ɵl)

7

*

1

( ) ( )L

MS BSBS MSl MS l BS l

l

N NH a a U V

L

……

Ba

se

ba

nd

Pre

co

de

r

Ba

se

ba

nd

Eq

ua

lize

r

……

RF

Be

am

form

erRF-Chain

RF-Chain

RF-Chain

RF-Chain

……

RF-Chain

RF-Chain

RF-Chain

RF-Chain

CSI

AcquisitionSpatially Sparse Precoding

SL- SVD

Tx Precoding for Hybrid Beamformer

RFF

MIMO

Channel

H

1V

BBF

RFW

BBW

AoD

H

…… ……

……

……

……

BBF

RFF

BSBS

BS 1a (θ )

BS

BS 2a (θ )

BS

BS 3a (θ )

MS

TdNjdj

BS

BS

BSlBS

BSl

leea ],...,,1[)(

)sin(2

)1()sin(2

)sin(2

)1(

)sin(2

3

3

1

BSBS

BS

dNj

dj

e

e

Page 8: Introduction to Hybrid Beamforming Techniquesaccess.ee.ntu.edu.tw/course/under_project_1032/... · Introduction to Hybrid Beamforming Techniques Graduate Institute of Electronics

ACCESS

Problem Formulation(3/3)

Step 2: Separate Fopt into(FBB ,FRF)

Due to spatial sparsity, this is

equivalent to solve an

optimization problem

Choose best Nrf columns to form FRF ,

and then Find FBB

8

BS

BS

BS 1a (θ )

BS

BS 2a (θ )

BS

BS 3a (θ )

MS

)sin(2

)1(

)sin(2

3

3

1

BSBS

BS

dNj

dj

e

e

st NNCV

1

LN

cantCA

sNL

BB CF

~

FBB

Nt: Number of Tx antennasNrf: Number of RF chainsL: Number of Active PathNs: Number of Tx data streams

FRF

……

Ba

se

ba

nd

Pre

co

de

r

Ba

se

ba

nd

Eq

ua

lize

r

……

RF

Be

am

form

erRF-Chain

RF-Chain

RF-Chain

RF-Chain

……

RF-Chain

RF-Chain

RF-Chain

RF-Chain

CSI

AcquisitionSpatially Sparse Precoding

SL- SVD

Tx Precoding for Hybrid Beamformer

RFF

MIMO

Channel

H

1V

BBF

RFW

BBW

AoD

H

…… ……

……

……

……

BBF

RFF

FRFFBB] )(a ,..., )(a , )(a , )(a[

1 T

BS

T

1BS

T

2BS

T

1BS

BS

L

BS

L

BSBS

tNAcan

FBBRFopt

BBRF

RFBB FFFFF

FF ,

minarg),(

Page 9: Introduction to Hybrid Beamforming Techniquesaccess.ee.ntu.edu.tw/course/under_project_1032/... · Introduction to Hybrid Beamforming Techniques Graduate Institute of Electronics

ACCESS

Existing Hybrid Beamforming Technique (I) (1/2)

[3] Use Orthogonal Matching Pursuit(OMP) to calculate (FBB ,FRF)

Perform Nrf iterations of correlation to find FRF

Perform pseudo-inverse to fine FBB

9

st NNCV

1

LN

cantCA

sNL

BB CF

~

FBB

Nt: Number of Tx antennasNrf: Number of RF chainsL: Number of Active PathNs: Number of Tx data streams

FRF

Page 10: Introduction to Hybrid Beamforming Techniquesaccess.ee.ntu.edu.tw/course/under_project_1032/... · Introduction to Hybrid Beamforming Techniques Graduate Institute of Electronics

ACCESS

Existing Hybrid Beamforming Technique (I) (2/2)

Hybrid precoding shows near optimal spatial efficiency

while compared with traditional baseband precoding

Spatial efficiency: the data rate that can be transmitted over a given bandwidth (units: bit/s/Hz)

Formula:

10

[3]

|)(|log *****1

2 BBRFRFBBBBRFRFBB

s

N WWHFFFHFWWRN

IRns

Page 11: Introduction to Hybrid Beamforming Techniquesaccess.ee.ntu.edu.tw/course/under_project_1032/... · Introduction to Hybrid Beamforming Techniques Graduate Institute of Electronics

ACCESS

Problem 1: Impractical Candidate Matrix

Impossible to get all AOD’s information

Require large bandwidth to return all AOD’s information from Rx

Need a candidate matrix without the information of All AOD

11

st NNCV

1

LN

cantCA

sNL

BB CF

~

FBB

Nt: Number of Tx antennasNrf: Number of RF chainsL: Number of Active PathNs: Number of Tx data streams

FRF

] )(a ,..., )(a , )(a , )(a[1

1 T

BS

T

1BS

T

2BS

T

1BS

BS

L

BS

L

BSBS

tNAcan

BSBS

BS 1a (θ )

BS

BS 2a (θ )

BS

BS 3a (θ )

MS

)sin(2

)1(

)sin(2

3

3

1

BSBS

BS

dNj

dj

e

e

Page 12: Introduction to Hybrid Beamforming Techniquesaccess.ee.ntu.edu.tw/course/under_project_1032/... · Introduction to Hybrid Beamforming Techniques Graduate Institute of Electronics

ACCESS

Problem 2: High ComplexityOptimization Algorithm

Long computation time for finding (FBB ,FRF)

OMP need Nrf iterations

Need an faster algorithm with less iterations

Pseudo-inverse is not suitable for HW implementation

Computational complexity:𝑂(𝑛3)

Need an algorithm without

pseudo-inverse

12

st NNCV

1

LN

cantCA

sNL

BB CF

~

FBB

Nt: Number of Tx antennasNrf: Number of RF chainsL: Number of Active PathNs: Number of Tx data streams

FRF

Page 13: Introduction to Hybrid Beamforming Techniquesaccess.ee.ntu.edu.tw/course/under_project_1032/... · Introduction to Hybrid Beamforming Techniques Graduate Institute of Electronics

ACCESS

Existing Hybrid Beamforming Technique (II) (1/3)

For problem 1, a DFT codebook is used

Predefined set: Consist of orthogonal column vectors

Don’t require all AOD’s information

Possibly find all Nrf columns using only 1 iteration

Equally space 360 degree with Nt angles to form

a full rank matrix

Hence Acan has Nt columns

13

st NNCV

1

tt NN

can CA

st NN

BB CF

~

Nt: Number of Tx antennasNrf: Number of RF chainsNs: Number of Tx data streams

FRF

FBB

Acan: DFT codebook

BS

BS

BS 1a (θ )

BS

BS 2a (θ )

BS

BS 3a (θ )

MS

)sin(2

)1(

)sin(2

3

3

1

BSBS

BS

dNj

dj

e

e

Page 14: Introduction to Hybrid Beamforming Techniquesaccess.ee.ntu.edu.tw/course/under_project_1032/... · Introduction to Hybrid Beamforming Techniques Graduate Institute of Electronics

ACCESS

Existing Hybrid Beamforming Technique (II) (2/3)

For problem 2, OBMP with DFT codebook is used instead of OMP with Acan1

Constraints: Acan must be orthogonal

Using 1 iteration to find (FBB ,FRF)

No pseudo-inverse

14

st NNCV

1

tt NN

can CA

st NN

BB CF

~

FRF

FBB

Algorithm : Othogonality-Based Matching Pursuit

optRequire : F

OPT1: F = Fres*

can res2: Ψ = A F*

,3: k = {n | n is the largest N index of ( ) }RF l l(k)

RF can4: F = A*

BB RF opt5: F = F FBB

BB s

opt RF BB

F6: F = N

F -F F

RF BB7: return F , F

Page 15: Introduction to Hybrid Beamforming Techniquesaccess.ee.ntu.edu.tw/course/under_project_1032/... · Introduction to Hybrid Beamforming Techniques Graduate Institute of Electronics

ACCESS

Existing Hybrid Beamforming Technique (II) (3/3)

OBMP’s computation time for finding (FBB ,FRF)

is less then that of OMP by 89.6% when Nrf equals 8

15

89.6%

Page 16: Introduction to Hybrid Beamforming Techniquesaccess.ee.ntu.edu.tw/course/under_project_1032/... · Introduction to Hybrid Beamforming Techniques Graduate Institute of Electronics

ACCESS

Summary

Advantage of hybrid beamforming

Reduce the number of RF chains but remain

near optimal performance

Design goal of hybrid beamforming

Method for finding (FBB ,FRF)

16

OMP[3] OBMP

Number of iteration Nrf 1

Complexity High Low

Constraints None OrthogonalAcan

FBBRFopt

BBRF

RFBB FFFFF

FF ,

minarg),(

Page 17: Introduction to Hybrid Beamforming Techniquesaccess.ee.ntu.edu.tw/course/under_project_1032/... · Introduction to Hybrid Beamforming Techniques Graduate Institute of Electronics

ACCESS

Reference

[1] M. Vu and A. Paulraj, “MIMO wireless linear precoding,” IEEE Signal Process. Mag., vol. 24, no. 5, pp. 86–105, Sept. 2007.

[2] Roh, W.; Ji-Yun Seol; Jeongho Park; Byunghwan Lee; Jaekon Lee; Yungsoo Kim; Jaeweon Cho; Kyungwhoon Cheun; Aryanfar, F., "Millimeter-wave beamforming as an enabling technology for 5G cellular communications: theoretical feasibility and prototype results," Communications Magazine, IEEE , vol.52, no.2, pp.106,113, February 2014

[3] El Ayach, O.; Rajagopal, S.; Abu-Surra, S.; Zhouyue Pi; Heath, R.W., "Spatially Sparse Precoding in Millimeter Wave MIMO Systems," Wireless Communications, IEEE Transactions on , vol.13, no.3, pp.1499,1513, March 2014

[4] D. P. Palomar, J. M. Cioffi, and M. A. Lagunas, “Joint Tx-Rx beamforming design for multicarrier MIMO channels: a unified framework for convex optimization,” IEEE Trans. Signal Process., vol. 51, no. 9, pp. 2381–2401, 2003.

17