47
Networking with massive MU- MIMO Lin Zhong http://recg.org

Networking with massive MU-MIMO

  • Upload
    newman

  • View
    108

  • Download
    0

Embed Size (px)

DESCRIPTION

Networking with massive MU-MIMO. Lin Zhong http:// recg.org. Guiding Principles. Spectrum is scarce Hardware is cheap, and getting cheaper. Antennas. Omni-directional base station. Data 1. Poor spatial reuse; poor power efficiency; high inter-cell interference. Sectored base station. - PowerPoint PPT Presentation

Citation preview

Page 1: Networking with  massive  MU-MIMO

Networking with massive MU-MIMO

Lin Zhonghttp://recg.org

Page 2: Networking with  massive  MU-MIMO

2

Guiding Principles

• Spectrum is scarce

• Hardware is cheap, and getting cheaper

Page 3: Networking with  massive  MU-MIMO

3

Antennas

Page 4: Networking with  massive  MU-MIMO

4

Data 1

Omni-directional base station

Poor spatial reuse; poor power efficiency; high inter-cell interference

Page 5: Networking with  massive  MU-MIMO

5

Data 1Data 2

Data 3

Sectored base station

Better spatial reuse; better power efficiency; high inter-cell interference

Page 6: Networking with  massive  MU-MIMO

6

Data 2

Data 1Data 3

Data 5

Single-user beamforming base station

Better spatial reuse; best power efficiency; reduced inter-cell interference

Page 7: Networking with  massive  MU-MIMO

7

Data 2

Data 1

Data 6

Data 3

Data 4Data 5

Multi-user MIMO base station

M: # of BS antennasK: # of clients (K ≤ M)

Best spatial reuse; best power efficiency; reduced inter-cell interference

Page 8: Networking with  massive  MU-MIMO

8

Key benefits of MU-MIMO• High spectral efficiency

• High energy efficiency

• Low inter-cell interference

• Orthogonal to Small Cell solutions– Centralized vs. distributed antennas

Page 9: Networking with  massive  MU-MIMO

9

Why massive?

• More antennas Higher spectral efficiency

• More antennas Higher energy efficiency

• Simple baseband technique becomes effective

T.L. Marzetta. Noncooperative cellular wireless with unlimited numbers of base station antennas. IEEE Trans. on Wireless Comm., 2010.

Page 10: Networking with  massive  MU-MIMO

Background: Beamforming

10

Page 11: Networking with  massive  MU-MIMO

Background: Beamforming

11

=

Constructive Interference

Page 12: Networking with  massive  MU-MIMO

Background: Beamforming

12

=

Constructive Interference

=

Destructive Interference

Page 13: Networking with  massive  MU-MIMO

Background: Beamforming

13

=

Destructive Interference

=

Constructive Interference

Page 14: Networking with  massive  MU-MIMO

Background: Beamforming

14

?

Page 15: Networking with  massive  MU-MIMO

15

Due to environment and terminal mobility estimation has to occur quickly and periodically

BS

The CSI is then calculated at the terminal and sent back to the BSA pilot is sent from each BS antenna

Background: Channel Estimation

+

+

Align the phases at the receiver to ensure constructive interferenceFor uplink, send a pilot from theterminal then calculate CSI at BS

Path Effects (Walls)Uplink?

Page 16: Networking with  massive  MU-MIMO

16

Background: Multi-user MIMO

BS

M: # of BS antennas K: # of clients

K ≤ M

Page 17: Networking with  massive  MU-MIMO

17

Multi-user MIMO: Precoding

BS

(M x 1 matrix)(Kx1 matrix)

M: # of BS antennas K: # of clients

K ≤ M

Page 18: Networking with  massive  MU-MIMO

18

Linear Precoding

BS

(M x 1 matrix)(Kx1 matrix)

M: # of BS antennas K: # of clients

K ≤ M

sWs

Page 19: Networking with  massive  MU-MIMO

19

Background: Zeroforcing Beamforming

Data 1

Null

Null

NullNullNull

Page 20: Networking with  massive  MU-MIMO

20

Background: Zeroforcing Beamforming

Data 2

Null

NullNull

Null

Null

Page 21: Networking with  massive  MU-MIMO

21

Background: Zeroforcing Beamforming

Data 2

Data 1

Data 6

Data 3

Data 4Data 5

Page 22: Networking with  massive  MU-MIMO

22

Background: Conjugate Beamforming

Data 1

Page 23: Networking with  massive  MU-MIMO

23

With more antennas

Data 1

Page 24: Networking with  massive  MU-MIMO

24

With even more antennas

Data 1

Page 25: Networking with  massive  MU-MIMO

Data 3

Data 5

Conjugate Multi-user Beamforming

Data 1

Data 6

Data 2

Data 4

Conjugate approaches Zeroforcing as M/K∞

Page 26: Networking with  massive  MU-MIMO

26

Conjugate vs. Zeroforcing

• Trivial computation

• Suboptimal capacity

• Scalable

• Nontrivial computation

• Close to capacity achieving

• Not scalable

Page 27: Networking with  massive  MU-MIMO

27

Recap

1) Estimate channels

2) Calculate weights

3) Apply linear precoding

Page 28: Networking with  massive  MU-MIMO

28

Scalability Challenges

1) Estimate channels– M+K pilots, then M•K feedback

2) Calculate weights– O(M•K2), non-parallelizable, centralized

data

3) Apply linear precoding– O(M•K), then O(M) data transport

Page 29: Networking with  massive  MU-MIMO

Argos’ Solutions

1) Estimate channels– New reciprocal calibration method

2) Calculate weights– Novel distributed beamforming method

3) Apply linear precoding– Carefully designed scalable architecture

O(M•K) → O(K)

O(M•K2) → O(K)

O(M•K) → O(K)

C. Shepard et al. Argos: Practical many-antenna base stations. ACM MobiCom, 2012.

Page 30: Networking with  massive  MU-MIMO

30

Solution: Argos ArchitectureCentral

Controller

Argos Hub

Argos Hub

Argos Hub

Module

Module

Module

Module

Module

Data Backhaul

ModuleRadio Radio Radio…

Page 31: Networking with  massive  MU-MIMO

31

Argos ImplementationCentral

Controller

Argos Hub

Module

Module

Module

…Central

Controller(PC with MATLAB)

Sync Pulse

Ethernet

Clock Distribution

Argos Hub Argos

Interconnect

WARP ModuleFPGA

Power PC

FPGA FabricHardware

ModelPeripherals and Other

I/O

Clock Board

Daughter

Cards

Radio 4

Radio 3

Radio 2

Radio 1

WARP ModuleFPGA

Power PC

FPGA FabricHardware

ModelPeripherals and Other

I/O

Clock Board

Daughter

Cards

Radio 4

Radio 3

Radio 2

Radio 1

16

………

Ethe

rne

t

WARP ModuleFPGA

Power PC

FPGA FabricHardware

ModelPeripherals and Other

I/O

Clock Board

Daughter

Cards

Radio 4

Radio 3

Radio 2

Radio 1

ArgosInterconne

ct

Page 32: Networking with  massive  MU-MIMO

32

Page 33: Networking with  massive  MU-MIMO

33

WARP Module

s

Central Controll

er

Argos Hub

Clock Distributio

nEtherne

t Switch

SyncDistributio

n

Argos Interconnects

Page 34: Networking with  massive  MU-MIMO

34

Experimental Setup• Time Division Duplex (TDD)

– Uplink and Downlink use the same band

• DownlinkListen to pilot

Calculate BF weights

Send data

Page 35: Networking with  massive  MU-MIMO

35

Conjugate vs. Zeroforcing

Page 36: Networking with  massive  MU-MIMO

36

Without considering computation

Listen to pilot

Calculate BF weights

Send data

Page 37: Networking with  massive  MU-MIMO

37

Linear gains as # of BS antennas increases

Capacity vs. M, with K = 15

Page 38: Networking with  massive  MU-MIMO

38

Linear gains as # of users increasesCapacity vs. K, with M = 64

Page 39: Networking with  massive  MU-MIMO

39

Considering computation

Listen to pilot

Calculate BF weights

Send data

Page 40: Networking with  massive  MU-MIMO

40Zeroforcing with various hardware configurations

M = 64 K = 15

Page 41: Networking with  massive  MU-MIMO

41

Conclusion• First many-antenna beamforming platform– Demonstration of manyfold capacity increase

• Devised novel techniques and architecture– Unlimited Scalability

• Simplistic conjugate beamforming works

• Need adaptive solutions

Page 42: Networking with  massive  MU-MIMO

42

Ongoing work

A network of massive MU-MIMO base stations

• Inter-cell interference management

• Pilot contamination• Client grouping & scheduling

Page 43: Networking with  massive  MU-MIMO

43

Page 44: Networking with  massive  MU-MIMO

44

~$2,000 per antenna

Page 45: Networking with  massive  MU-MIMO

45

Acknowledgments

http://argos.rice.edu

Page 46: Networking with  massive  MU-MIMO

More BS antennas + MU-MIMO Higher efficiency & lower interference

Data 2

Data 1

Data 6

Data 3

Data 4Data 5

Page 47: Networking with  massive  MU-MIMO

Data 10

Data 12

Data 2

Data 8

Data 6

Dat

a 4

Data 5

Data 9

Data 1

Data 11

Data 3

Data 7

More BS antennas + MU-MIMO Higher efficiency & lower interference