Centre for Communications Research Power Efficient MIMO Techniques for 3GPP LTE and Beyond K. C. Beh, C. Han, M. Nicolaou, S. Armour, A. Doufexi.

Centre for Communications Research

Power Efficient MIMO Techniques for 3GPP LTE and BeyondK. C. Beh, C. Han, M. Nicolaou, S. Armour, A. Doufexi

Green Radio

• 4 billion mobile phone users worldwide

• Telecommunication industry responsible for 183 million tons of CO2

• MVCE framework (Core 5): Deliver high data rate services with a 100-fold reduction in power consumption

Green Radio and LTE

• LTE next major step in mobile radio communications

• Aim to reduce delays, improve spectrum flexibility, reduce cost of operators and end users

• MIMO transmission techniques improve system reliability and performance

• LTE support of a MIMO scheduling and precoding method with

improved interface between PHY and DLC

Green Radio and LTE

• Examine performance of proposed MIMO-OFDMA scheme

• Consider the capabilities of MIMO-OFDMA precoding in reducing Tx. Power from Base Station (BS)

• Theoretical analysis and simulation results

• Maintain QoS levels with reduced Tx. Power

System and Channel Model• Spatial Channel Model Extension (SCME) Urban Macro

• Low spatially correlated channel for all users

• 2x2 MIMO architecture (analysis is readily extendible to higher MIMO orders)

• Perfect CQI estimation and feedback

• Ideal Link Adaptation based on 6 Modulation and Coding Schemes (MCS)

System and Channel ModelTransmission Bandwidth 10 MHz

Time Slot/Sub-frame duration 0.5ms/1ms

Sub-carrier spacing 15kHz

Sampling frequency 15.36MHz (4x3.84MHz)

FFT size 1024

Number of occupiedsub-carriers

600

Number of OFDM symbols per time slot (Short/Long CP)

7/6

CP length (μs/samples)

Short (4.69/72)x6(5.21/80)x1

Long (16.67/256)

System and Channel ModelMode Modulation Cod. Rate Data bits per time slot

(1x1), (2x2)Bit Rate(Mbps)

1 QPSK 1/2 4000/7600 8/15.2

2 QPSK 3/4 6000/11400 12/22.8

3 16 QAM 1/2 8000/15200 16/30.4

4 16 QAM 3/4 12000/22800 24/45.6

5 64 QAM 1/2 12000/22800 24/45.6

6 64 QAM 3/4 18000/34200 36/68.4

Random and Layered Random Beamforming

• Random Unitary Matrix applied to frequency sub-carriers on Physical Resource Block (PRB) basis

• Linear MMSE Receiver with interference suppression capability• MIMO channels can be decomposed into separate spatial layers• ESINR feedback for resource allocation• Random Beamforming: All spatial layers to a single user• Layered Random Beamforming: Spatial layers assigned to different

users Higher Diversity

Unitary Codebook Based Beamforming

• Pre-defined set of antenna beams• Pre-coders based on Fourier basis for uniform sector coverage• Variable codebook size G, consisting of the unitary matrix set• Large Codebook: Higher Spatial Granularity, Increased Feedback• Small Codebook: Low Spatial Granularity, Lower Feedback• Single-User MIMO (SU-MIMO) and Multi-User MIMO (MU-MIMO)

capability

Feedback Considerations• Full Feedback: CQI for all

precoding matrices• Partial Feedback: CQI on

preferred beams• Suboptimal performance for MU-

MIMO with partial feedback• Codebook size G=2 assumed

Theoretical Analysis• Precoding schemes achieve varying degrees of Multiuser

Diversity (MUD) (K=5)• A target spectral efficiency achieved at different SNR levels for

different schemes

-4 -2 0 2 4 6 8 10 12 14 160

2

4

6

8

10

12

SNR (dB)

Spe

ctra

l Effi

cien

cy (

bps/

Hz)

SISOSFBC

Random Beamforming

Layered Random Beamforming

SU-MIMO

Full Feedb. MU-MIMOPartial Feedb. MU-MIMO

Theoretical Analysis• Target Spectral Efficiency 3bps/Hz• Single User SISO Benchmark• Higher benefits for increasing numbers of users• K=10, MU-MIMO, Gain= 5dB

1 5 10 15 20 250

2

4

6

8

10

12

No.of User

Per

form

ance

Gai

n O

ver

Sin

gle

Use

r S

ISO

(dB

)

SISOSFBC

Random Beamforming

Layered Random Beamforming

SU-MIMO

Full Feedb. MU-MIMOPartial Feedb. MU-MIMO

Simulation Results

• Analysis based on ideal Adaptive Modulation and Coding (AMC)

• Throughput = R(1-PER),• Results consistent with

theoretical analysis

-10 -5 0 5 10 15 20 250

1

2

3

4

5

6

7

SNR (dB)

Spe

ctra

l Eff

icie

ncy

(bps

/Hz)

SISOSFBC

Random Beamforming

Layered Random Beamforming

SU-MIMO

Full Feedb. MU-MIMOPartial Feedb. MU-MIMO

Simulation Results• Simulation performance predicts

even higher gains • Actual performance PER

dependent. • MU-MIMO and LRB eliminate deep

fades that cause severe link degradations

• MU-MIMO gain @ K=10: 7dB • SFBC suffers from inherent inability

to exploit MUD

1 5 10 15 20 250

5

10

15

No.of UserP

erfo

rman

ce G

ain

Ove

r S

ingl

e U

ser

SIS

O (

dB)

SISOSFBC

Random Beamforming

Layered Random Beamforming

SU-MIMO

Full Feedb. MU-MIMOPartial Feedb. MU-MIMO

Power Efficiency and Fairness• Power Efficiency associated with a cost metric

and a corresponding Power Fairness Index (PFI)

• Low cost metric implies high power efficiency

2

1

2

1

K

kk

kK

lk

k

R

PK

R

PPFI

K

kkk RP

1

Cost Metric Variance

SISO 2.1593 0.6923

Random Beamforming 0.9171 0.0573

Layered Random Beamforming 0.8699 0.0214

SU-MIMO 0.9172 0.0575

Full Feedb. MU-MIMO 0.8536 0.0285

Partial Feedb. MU-MIMO 0.8602 0.0312

Power Efficiency and Fairness

• PFI indication of how fairly power is allocated to different users with respect to their achieved rates

• Uplink improvements• Schemes utilising the additional

spatial layer, achieve an overall higher power allocation fairness, with PFI values consistently closer to unity.

Conclusions and Future Work• Multiuser Diversity schemes exploiting temporal, spectral and

spatial domain achieve notable performance gains. • Performance gains can be translated to a power saving at the BS• Theoretical Analysis consistent with simulation results• Improved consistency in cost metric• Improved power allocation fairness• Power savings of up to 10dB can be achieved with no QoS

compromise