22.01.2014 Views

Advances in Coordinated Multi-Cell Multi-User MIMO Systems

Advances in Coordinated Multi-Cell Multi-User MIMO Systems

Advances in Coordinated Multi-Cell Multi-User MIMO Systems

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

<strong>Advances</strong> <strong>in</strong> Coord<strong>in</strong>ated <strong>Multi</strong>-<strong>Cell</strong> <strong>Multi</strong>-<strong>User</strong> <strong>MIMO</strong><br />

<strong>Systems</strong><br />

Li-Chun Wang <br />

Department of Electrical Eng<strong>in</strong>eer<strong>in</strong>g<br />

National Chiao Tung University<br />

Hs<strong>in</strong>chu, Taiwan<br />

http://lichun.cm.nctu.edu.tw<br />

lichun@cc.nctu.edu.tw


Information for presentation slides<br />

• http://www.ieee-globecom.org/2011/private


Outl<strong>in</strong>e<br />

1 Introduction to <strong>MIMO</strong> Antenna<br />

Techniques<br />

2 S<strong>in</strong>gle-cell <strong>Multi</strong>-user <strong>MIMO</strong> <strong>Systems</strong><br />

3 S<strong>in</strong>gle-cell <strong>Multi</strong>-user <strong>MIMO</strong> Broadcast<br />

<strong>Systems</strong><br />

4 <strong>Multi</strong>-cell Network <strong>MIMO</strong> systems<br />

5 Updates of Network <strong>MIMO</strong> Techniques <strong>in</strong><br />

3GPP LTE-A Updates<br />

6 Conclusions


Ma<strong>in</strong> Story<br />

• Story 1: Schedul<strong>in</strong>g <strong>in</strong> diversity-based <strong>MIMO</strong><br />

• Story 2: Schedul<strong>in</strong>g for multiplex<strong>in</strong>g-based <strong>MIMO</strong><br />

• Story 3: A robust multi-user <strong>MIMO</strong> broadcast<br />

system aga<strong>in</strong>st huge feedback channel<br />

variations<br />

• Story 4: Channel assignment for multi-cell<br />

Network <strong>MIMO</strong><br />

• Story 5: Viewpo<strong>in</strong>ts of 3GPP on network <strong>MIMO</strong>


What happened <strong>in</strong> 1895?


1895 <strong>in</strong> Taiwan


1895 <strong>in</strong> Italy<br />

It is dangerous to put limits on wireless! (1932)


Guglielmo Marconi (1874 ~ 1937)<br />

• In 1895 he began laboratory experiments and<br />

succeeded <strong>in</strong> send<strong>in</strong>g wireless signals over a<br />

distance of one and a half miles.<br />

• In 1896, he was granted the world's first patent for a<br />

system of wireless telegraphy.<br />

• On 12 December 1901, us<strong>in</strong>g a 150 meter (500ft) kitesupported<br />

antenna for reception, the message fwas<br />

received at Signal Hill <strong>in</strong> Newfoundland of Canada by the<br />

company's new high-power station at Poldhu, which was<br />

transmitted about 3,500 kilometres (2,200 mi).<br />

• Guglielmo Marconi and Ferd<strong>in</strong>and Braun won the<br />

Nobel Prize <strong>in</strong> Physics 1909.


Wireless 101<br />

The more th<strong>in</strong>gs changes, the more rema<strong>in</strong> the same.<br />

(Alphonse Karr, 1808)


What is the fundamental issue for wireless<br />

communications?<br />

Spectrum Efficiency


Some Milestones <strong>in</strong> Telecommunications<br />

• Fundamental resources (degrees of freedom) <strong>in</strong><br />

communication system eng<strong>in</strong>eer<strong>in</strong>g<br />

– Power (pre-1948),<br />

– Bandwidth (1948, Shannon Capacity)<br />

– complexity (1980, TCM),<br />

– Space (1998, Space-Times Process<strong>in</strong>g )<br />

11


An Open Issue<br />

What is the next possible degree of freedom<br />

that can be exploited for communications<br />

systems?<br />

12


Research Trends <strong>in</strong> Wireless Networks<br />

• The Past Two Decades: Key Developments at the L<strong>in</strong>k<br />

Level : <br />

– <strong>MIMO</strong>; MUD; Turbo<br />

• Today: An Increased Focus on Interactions Among Nodes<br />

– Competition <br />

» Cognitive radio <br />

» Information theoretic security <br />

» Game theoretic model<strong>in</strong>g, analysis & design<br />

– Collaboration <br />

» Network cod<strong>in</strong>g <br />

» Cooperative transmission & relay<strong>in</strong>g <br />

» <strong>Multi</strong>-hop transmission & coalition games <br />

» Collaborative beam-form<strong>in</strong>g <br />

» Collaborative <strong>in</strong>ference <br />

– Competition & Collaboration <strong>in</strong> Wireless Networks


H<strong>in</strong>t from this Talk<br />

How can we effectively exploit the degree of<br />

freedom <strong>in</strong> the user doma<strong>in</strong> to design multiuser<br />

<strong>MIMO</strong> and multi-cell network <strong>MIMO</strong><br />

systems?<br />

14


1. Introduction to <strong>MIMO</strong> Antenna<br />

Techniques<br />

(Story 1: Schedul<strong>in</strong>g <strong>in</strong> diversity-based<br />

<strong>MIMO</strong>)


<strong>Multi</strong>-Input <strong>Multi</strong>-Output (<strong>MIMO</strong>)<br />

Antenna Techniques<br />

• Independent parallel transmit and receive antenna<br />

pair<br />

• <strong>MIMO</strong> antenna techniques<br />

– boost channel capacity<br />

– enhance l<strong>in</strong>k reliability<br />

– reject strong <strong>in</strong>terference<br />

N t<br />

N r<br />

Tx<br />

…<br />

…<br />

Rx<br />

16


Diversity-<strong>Multi</strong>plex<strong>in</strong>g Tradeoff<br />

• Diversity -> improve l<strong>in</strong>k reliability by replicas<br />

a<br />

Tx<br />

a<br />

a<br />

a<br />

a<br />

Rx<br />

a<br />

• Multuplex<strong>in</strong>g -> enhance data rate by multiplex<strong>in</strong>g<br />

ba<br />

Tx<br />

a<br />

b<br />

a+b<br />

b+a<br />

Rx<br />

ba<br />

subchannel <strong>in</strong>terference<br />

cancel<strong>in</strong>g is required<br />

17


Schedul<strong>in</strong>g Techniques<br />

• Through periodically select<strong>in</strong>g the best user to serve, the<br />

system performance is improved by exploit<strong>in</strong>g multiuser<br />

diversity or cooperative diversity.<br />

• Ordered Statistics is the fundamental mathematical<br />

techniques for analyz<strong>in</strong>g the schedul<strong>in</strong>g wireless systems.<br />

user 1<br />

user 1<br />

user 2<br />

user 3<br />

1 2 3 4 5 6 7 8 9<br />

1 2 3 4 5 6 7 8 9<br />

18


A <strong>Multi</strong>user <strong>MIMO</strong> Schedul<strong>in</strong>g System<br />

• Perfect SNR estimation and noiseless feedback<br />

• H k =[ h ij<br />

(k)<br />

], each h ij<br />

(k)<br />

is subject to Nakagami fad<strong>in</strong>g<br />

• The BS selects the target user with highest effective<br />

SNR<br />

19


The Benefit from <strong>Multi</strong>user Schedul<strong>in</strong>g<br />

multiuser diversity or cooperative diversity


Nakagami-m Fad<strong>in</strong>g Channel <br />

Which one delivers the highest capacity ?


Impact of Channel Fad<strong>in</strong>g on System Capacity<br />

m=1<br />

m=1<br />

Two totally different stories for K=1 and K >1


A <strong>Multi</strong>user <strong>MIMO</strong> Schedul<strong>in</strong>g System<br />

• Perfect SNR estimation and noiseless feedback<br />

• H k =[ h ij<br />

(k)<br />

], each h ij<br />

(k)<br />

is subject to Nakagami fad<strong>in</strong>g<br />

• The BS selects the target user with highest effective SNR


A Generic Diversity-Based <strong>MIMO</strong> System<br />

The maximum diversity ga<strong>in</strong> = N t N r


Some Diversity-Based <strong>MIMO</strong> Schemes<br />

• Four exist<strong>in</strong>g diversity-based <strong>MIMO</strong> schemes are<br />

considered<br />

– ST/SC<br />

– ST/MRC<br />

– MRT/MRC<br />

– STBC<br />

ST : selective transmission<br />

SC : selective comb<strong>in</strong><strong>in</strong>g<br />

MRT : maximum ratio transmission<br />

MRC : maximum ratio comb<strong>in</strong><strong>in</strong>g<br />

STBC : space time block code<br />

• All of them can deliver full antenna diversity ga<strong>in</strong>


When Diversity-Based <strong>MIMO</strong> Meets<br />

<strong>Multi</strong>user Schedul<strong>in</strong>g<br />

• In general, schedul<strong>in</strong>g is a MAC layer technique to<br />

deliver multiuser diversity ga<strong>in</strong> by exploit<strong>in</strong>g<br />

<strong>in</strong>dependent channel fluctuations among users<br />

• By contrast, antenna diversity is a physical layer<br />

approach to offer reliable transmissions with the<br />

major goal of mitigat<strong>in</strong>g channel fad<strong>in</strong>g<br />

Induced l<strong>in</strong>k variation by<br />

diversity-based <strong>MIMO</strong><br />

SISO


A Closer Look for their Interplay<br />

Diversity-based<br />

<strong>MIMO</strong> L<strong>in</strong>k<br />

“peak”<br />

AF<br />

Array ga<strong>in</strong><br />

AF<br />

SISO


An Analytical Upper Bound<br />

• In [Chen & Wang ’04], we show that<br />

system capacity<br />

selection order<br />

AF ga<strong>in</strong><br />

array ga<strong>in</strong><br />

SNR<br />

Nakagami fad<strong>in</strong>g parameter


ST/SC over <strong>Multi</strong>user Schedul<strong>in</strong>g System<br />

SISO<br />

ST/SC<br />

MS 1<br />

MS 1<br />

virtual users<br />

MS 4<br />

virtual antennas<br />

MS 2<br />

MS 2<br />

MS 5<br />

Tx<br />

Rx-<br />

SC<br />

MS 3<br />

K=3, N t =1, N r =2<br />

MS 3<br />

MS6<br />

K=6, N t =1, N r =1<br />

K=1, N t =1, N r =6


MRT/MRC over <strong>Multi</strong>user Schedul<strong>in</strong>g System<br />

SISO<br />

SC<br />

MRC<br />

MRC<br />

SISO


STBC over <strong>Multi</strong>user Schedul<strong>in</strong>g System<br />

STBC (N t ,1)<br />

SISO<br />

STBC<br />

channel damp<strong>in</strong>g<br />

Fad<strong>in</strong>g<br />

channel<br />

m<br />

MS 1<br />

MS 2<br />

=<br />

Fad<strong>in</strong>g<br />

channel<br />

4*m<br />

MS 1<br />

MS 2<br />

<<br />

Fad<strong>in</strong>g<br />

channel<br />

m<br />

MS 1<br />

MS 2


System Capacity with Jo<strong>in</strong>t Antenna<br />

and <strong>Multi</strong>user Diversity


Ma<strong>in</strong> Po<strong>in</strong>t <strong>in</strong> Story 1<br />

• <strong>Multi</strong>user schedul<strong>in</strong>g and the diversity-based <strong>MIMO</strong><br />

with the multiuser schedul<strong>in</strong>g system may not be a<br />

good marriage.<br />

• Why?<br />

– <strong>User</strong> population, i.e. K, has contributed a large<br />

amount of diversity <strong>in</strong> the whole system<br />

– More important is their <strong>in</strong>tr<strong>in</strong>sic conflicts – one<br />

prefers variation, while the other creates<br />

tranquility


2. S<strong>in</strong>gle-cell <strong>Multi</strong>-user <strong>MIMO</strong><br />

<strong>Systems</strong><br />

(Story 2: Schedul<strong>in</strong>g for multiplex<strong>in</strong>g-based<br />

<strong>MIMO</strong>)


Issue for a Spatial <strong>Multi</strong>plex<strong>in</strong>g-based <strong>MIMO</strong><br />

• Diversity-multiplex<strong>in</strong>g tradeoff <strong>in</strong> a po<strong>in</strong>t-to-po<strong>in</strong>t <strong>MIMO</strong> system<br />

– multiplex<strong>in</strong>g ga<strong>in</strong> comes at the price of diversity ga<strong>in</strong> [Zheng &<br />

Tse ’03]<br />

– may translate <strong>in</strong>to smaller coverage areas if the SM <strong>MIMO</strong><br />

scheme is used [Catreux & Greenste<strong>in</strong> ‘03]<br />

MS<br />

• The coverage is def<strong>in</strong>ed as the maximum distance at which<br />

the l<strong>in</strong>k suffices for ma<strong>in</strong>ta<strong>in</strong><strong>in</strong>g a required receive SNR γ th with<br />

a probability, say 90%, at least<br />

35


Schedul<strong>in</strong>g Techniques<br />

• Through periodically select<strong>in</strong>g the best user to serve, the<br />

system performance is improved by exploit<strong>in</strong>g multiuser<br />

diversity or cooperative diversity.<br />

• Ordered Statistics is the fundamental mathematical<br />

techniques for analyz<strong>in</strong>g the schedul<strong>in</strong>g wireless systems.<br />

user 1<br />

user 1<br />

user 2<br />

user 3<br />

1 2 3 4 5 6 7 8 9<br />

1 2 3 4 5 6 7 8 9<br />

36


A <strong>Multi</strong>user <strong>MIMO</strong> Schedul<strong>in</strong>g System<br />

• Perfect SNR estimation and noiseless feedback<br />

• H k =[ h ij<br />

(k)<br />

], each h ij<br />

(k)<br />

is subject to Nakagami fad<strong>in</strong>g<br />

• The BS selects the target user with highest effective<br />

SNR<br />

37


The SWNSF Schedul<strong>in</strong>g Game<br />

SWNSF: strongest-weakestnormalized-subchannel-first<br />

<br />

MS 1<br />

MS 2<br />

MS 3<br />

38


Effect of SWNSF Schedul<strong>in</strong>g on m<strong>in</strong><br />

SWNSF schedul<strong>in</strong>g enhances the output SNR of the weakest subchannel<br />

λ 3 > λ 2 >λ 1


s<br />

u<br />

i<br />

d<br />

a<br />

r<br />

Effect of SWNSF Schedul<strong>in</strong>g on Coverage<br />

e<br />

g<br />

a<br />

r<br />

e<br />

v<br />

o<br />

c<br />

d<br />

e<br />

z<br />

i<br />

l<br />

a<br />

m<br />

r<br />

o<br />

N<br />

1.8<br />

1.6<br />

1.4<br />

1.2<br />

1<br />

0.8<br />

0.6<br />

N=2<br />

N=3<br />

MS<br />

<strong>MIMO</strong> + RR Schedul<strong>in</strong>g<br />

MS<br />

<strong>MIMO</strong> + SWNSF Schedul<strong>in</strong>g<br />

(“soft” coverage depends on K)<br />

0.4<br />

1 5 10 15 20 25 30<br />

Number of users K<br />

40


Effect of SWNSF Schedul<strong>in</strong>g on Other i


Coverage Extension & Capacity Improvement<br />

number of users<br />

number of antennas<br />

43


Ma<strong>in</strong> Po<strong>in</strong>t <strong>in</strong> Story 2<br />

<strong>Multi</strong>-<strong>User</strong> Schedul<strong>in</strong>g is<br />

A Soft Coverage Extension Technique<br />

"Enhanc<strong>in</strong>g Coverage and Capacity for <strong>Multi</strong>user <strong>MIMO</strong> <strong>Systems</strong> by Utiliz<strong>in</strong>g Schedul<strong>in</strong>g, "<br />

IEEE Trans. on Wireless Communications, Vol. 5, No. 5, pp. 1148-1157, May, 2006.


3. S<strong>in</strong>gle-cell <strong>Multi</strong>-user <strong>MIMO</strong><br />

Broadcast <strong>Systems</strong><br />

Story 3: A robust multi-user <strong>MIMO</strong> broadcast system<br />

aga<strong>in</strong>st huge feedback channel variations


<strong>MIMO</strong> Broadcast <strong>Systems</strong><br />

• <strong>MIMO</strong> antenna <strong>in</strong> po<strong>in</strong>t-to-po<strong>in</strong>t scenario can boost<br />

capacity through spatial multiplex<strong>in</strong>g.<br />

• <strong>MIMO</strong> broadcast systems (po<strong>in</strong>t-to-multiple) can transmit<br />

personalized data services to multiple users concurrently.<br />

Traditional po<strong>in</strong>t-to-po<strong>in</strong>t transmission<br />

Po<strong>in</strong>t-to-multiple transmissions <strong>in</strong> spatial doma<strong>in</strong>:<br />

<strong>Multi</strong>-<strong>User</strong> <strong>MIMO</strong> broadcast systems


Challenges <strong>in</strong> Transmit Beamform<strong>in</strong>g MU-<strong>MIMO</strong> <br />

• Types of MU-<strong>MIMO</strong> Transmit beamform<strong>in</strong>g systems:<br />

– Zero-forc<strong>in</strong>g (ZF) based transmit beamform<strong>in</strong>g<br />

– Block diagonalization (BD) based transmit<br />

beamform<strong>in</strong>g<br />

– Dirty paper cod<strong>in</strong>g (DPC)<br />

• Challenges of feedback channel design <strong>in</strong> transmit<br />

beamform<strong>in</strong>g MU-<strong>MIMO</strong> broadcast<br />

» Accuracy<br />

» Bandwidth => codebook solution


<strong>MIMO</strong> Broadcast <strong>Systems</strong> with<br />

Transmit Beamform<strong>in</strong>g<br />

H 1<br />

Receive<br />

Beamform<strong>in</strong>g<br />

Transmit<br />

Beamform<strong>in</strong>g<br />

W<br />

H K<br />

Receive<br />

Beamform<strong>in</strong>g<br />

H k<br />

M T M R complex-valued entries of channel matrix per user<br />

• Utilize feedback <strong>in</strong>formation to calculate the beamform<strong>in</strong>g weight W<br />

for <strong>in</strong>terference cancellation<br />

• ZF-DPC: QR-decomposition based transmit beamform<strong>in</strong>g<br />

• ZF: channel <strong>in</strong>verse based transmit beamform<strong>in</strong>g


<strong>MIMO</strong> Broadcast <strong>Systems</strong> with<br />

Jo<strong>in</strong>t Transmit Precod<strong>in</strong>g and Receive Equalizer<br />

Transmit<br />

Precod<strong>in</strong>g<br />

T k<br />

H 1<br />

Receive<br />

Equalizer<br />

Receive<br />

Equalizer<br />

H K<br />

R 1<br />

H k<br />

R K<br />

R k<br />

M T M R complex-valued entries of channel matrix per user<br />

Need return R k to each user!!<br />

• Utilize feedback <strong>in</strong>formation to calculate T k and R k for <strong>in</strong>terference<br />

cancellation – Block diagonalization (BD)


A Nonconformist (Why Not) Issue<br />

• Most MU-<strong>MIMO</strong> systems design transmit beamform<strong>in</strong>g<br />

(precod<strong>in</strong>g) at the transmitter side to cancel the <strong>in</strong>ter-stream<br />

<strong>in</strong>terference among serv<strong>in</strong>g users, but suffer from feedback<br />

channel errors and bandwidth issues.<br />

• Q. Can we have a different architecture to realize MU-<br />

<strong>MIMO</strong> broadcast systems?


Our Answer<br />

• Receive ZF beamform<strong>in</strong>g <strong>MIMO</strong> broadcast systems comb<strong>in</strong>ed<br />

with multiuser schedul<strong>in</strong>g<br />

• An alternative MU-<strong>MIMO</strong> technique to use feedback<br />

channel for a different purpose => user selection.<br />

• Has advantages <strong>in</strong> terms of robustness aga<strong>in</strong>st CSI<br />

and feedback l<strong>in</strong>k errors [Wang’10]<br />

[Wang’10] Li-Chun Wang and Chu-Jung Yeh, "Schedul<strong>in</strong>g for multiuser <strong>MIMO</strong><br />

broadcast systems: transmit or receive beamform<strong>in</strong>g?" IEEE Transactions on<br />

Wireless Communications, vol. 9, no. 9, pp. 2779 – 2791, Sep, 2010.


New Perspective: <strong>MIMO</strong> Broadcast <strong>Systems</strong> with<br />

Receive Beamform<strong>in</strong>g<br />

H 1<br />

Receive<br />

Beamform<strong>in</strong>g<br />

Transmit<br />

Beamform<strong>in</strong>g<br />

W 1<br />

[γ 1 , γ 2 , … ]<br />

M T SNR values<br />

H K<br />

Receive<br />

Beamform<strong>in</strong>g<br />

W K<br />

M T real-valued scalars per user (vector feedback)<br />

• Utilize feedback <strong>in</strong>formation to select the appropriate users<br />

• The receive ZF <strong>MIMO</strong> broadcast systems: channel <strong>in</strong>verse based<br />

receive beamform<strong>in</strong>g


Example of Receive ZF Beamform<strong>in</strong>g (3 x 3)<br />

• Feedback <strong>in</strong>formation<br />

– <strong>User</strong> 1: [1.25, 0.49, 0.50]<br />

– <strong>User</strong> 2: [0.81, 0.70, 2.25]<br />

– <strong>User</strong> 3: [0.49, 1.69, 1.21]<br />

Select<br />

Antenna 1 Antenna 2 Antenna 3<br />

<strong>User</strong> 1 <strong>User</strong> 3 <strong>User</strong> 2<br />

(Vector feedback)


Literature Survey on MU-<strong>MIMO</strong><br />

Tx BD Rx Note<br />

BD: Block diagonalization<br />

[Caire’03] o x x Concept of ZF-DPC and ZF beamform<strong>in</strong>g<br />

[J<strong>in</strong>dal’05] [Sharif’07] o x x The ga<strong>in</strong> of broadcast system over po<strong>in</strong>t-to-po<strong>in</strong>t<br />

transmission<br />

[Vishwanath’03]<br />

[Viswanath’03]<br />

[We<strong>in</strong>garten‘06]<br />

[Tu’03][Dimi´c’05]<br />

[Yoo’06][Bayesteh’08]<br />

[J<strong>in</strong>dal’06][Yoo’07]<br />

[Swannack‘06]<br />

[Caire’10] [D<strong>in</strong>g’07]<br />

[Zhang’09]<br />

[Spencer’04][Chen’08]<br />

[Shen’06,07]<br />

[Heath’01][Airy’04]<br />

[Chen’07]<br />

Our work <strong>in</strong> [Wang’10]<br />

[Wang’11a]<br />

[Wang’11b]<br />

o x x Theoretical capacity region of <strong>MIMO</strong> broadcast<br />

systems<br />

o x x Schedul<strong>in</strong>g algorithms and sum-rate performance<br />

evaluation and analysis<br />

o x x Under limited feedback condition: sum-rate analysis,<br />

codebook and schedul<strong>in</strong>g algorithm design<br />

x o x The concept, theoretical capacity, schedul<strong>in</strong>g<br />

algorithms of BD<br />

x x o Concept. scal<strong>in</strong>g law. sum-rate analysis under equal<br />

power allocation (M R = M T case)<br />

o o <br />

Effects of feedback channel variations (Part I)<br />

L<strong>in</strong>k performance analysis (Part 2)<br />

<br />

Effects of channel estimation error for receive ZF<br />

beamform<strong>in</strong>g (Part 3)


Key Analytic Results (M R = M T = M)<br />

• The sum rate of the receive ZF <strong>MIMO</strong> broadcast<br />

systems with water-fill<strong>in</strong>g power allocation<br />

The water-level solution of the equation<br />

– Utilize order statistics technique and long-term<br />

power constra<strong>in</strong>t for water-fill<strong>in</strong>g equation to derive<br />

the closed form<br />

Γ R (a,x): upper <strong>in</strong>complete gamma function<br />

E i (x): exponential <strong>in</strong>teger function of order i


Example of an MU-<strong>MIMO</strong> System with Receive ZF<br />

Beamform<strong>in</strong>g (3 x 3)<br />

• Feedback <strong>in</strong>formation<br />

– <strong>User</strong> 1: [1.25, 0.49, 0.50]<br />

– <strong>User</strong> 2: [0.81, 0.70, 2.25]<br />

– <strong>User</strong> 3: [0.49, 1.69, 1.21]<br />

Select<br />

Antenna 1 Antenna 2 Antenna 3<br />

<strong>User</strong> 1 <strong>User</strong> 3 <strong>User</strong> 2<br />

(Vector feedback)


Key Analytic Results (M R ¸ M T )<br />

• A general form of the sum rate with water-fill<strong>in</strong>g<br />

power allocation<br />

L = M R – M T<br />

Key steps <strong>in</strong> the closed form<br />

• Order statistics technique<br />

• Long-term power constra<strong>in</strong>t for water-fill<strong>in</strong>g<br />

equation<br />

• The <strong>in</strong>tegral identity provided <strong>in</strong> [Alou<strong>in</strong>i’99]<br />

The water-level solution of the equation


Sum Rate Comparison<br />

• Under similar feedback requirement<br />

M T values for feedback!!<br />

Transmit beamform<strong>in</strong>g: M T = {2, 4}; M R =1<br />

Receive beamform<strong>in</strong>g: M T = {2, 4} = M R<br />

RZFSTZFS <strong>in</strong> sum rate


What happen if feedback <strong>in</strong>formation is perturbed<br />

or channel estimation is <strong>in</strong>accurate?


Two Issues on channel <strong>in</strong>formation accuracy<br />

• Can receive beamform<strong>in</strong>g ZF <strong>MIMO</strong> broadcast systems<br />

tolerate feedback channel variations?<br />

• Can receive beamform<strong>in</strong>g ZF <strong>MIMO</strong> broadcast systems<br />

channel estimation errors?<br />

H 1<br />

Receive<br />

Beamform<strong>in</strong>g<br />

Transmit<br />

Beamform<strong>in</strong>g<br />

channel estimation errors<br />

[γ 1 , γ 2 , … ]<br />

H K<br />

Receive<br />

Beamform<strong>in</strong>g<br />

[Wang’07] each entry of<br />

E k is distributed by<br />

feedback channel variations (Part I-1)


Feedback CSI Variations<br />

• Adopt coefficient of variation (CV) to evaluate the impact<br />

of feedback CSI variations<br />

– Perturbed from the noisy estimated <strong>in</strong>formation, the<br />

outdated <strong>in</strong>formation, and quantization errors, etc.<br />

– For random variable X, CV = σ x / E[X]<br />

• For the feedback CSI x, the perturbed feedback CSI<br />

received at BS is<br />

x’ = x + <br />

– ~ N(0, σ x2 ) and σ x = CVx


Effects of Feedback CSI Variations<br />

on Sum Rate (CV = 0.5)<br />

K = 20, M T = M R = 3<br />

• With perfect CSI feedback<br />

– transmit beamform<strong>in</strong>g is better<br />

than receive beamform<strong>in</strong>g<br />

(1 ~ 1.5 nats/s/Hz)<br />

– ZF-DPC is the best one<br />

• With perturbed CSI feedback<br />

– Large sum rate degradation<br />

for transmit beamform<strong>in</strong>g<br />

(Reduce 19.7% ~ 33.8%)<br />

– Slight sum rate degradation for<br />

receive ZF beamform<strong>in</strong>g<br />

(Reduce 3.5%)


Effects of Feedback CSI Variations<br />

on Sum Rate (CV = 1.5)<br />

K = 20, M T = M R = 3<br />

• Receive ZF beamform<strong>in</strong>g is<br />

more robust to feedback CSI<br />

variations<br />

– 7% sum rate reduction<br />

– Keep the sum rate slope<br />

• Transmit beamform<strong>in</strong>g is more<br />

sensitive to feedback CSI<br />

variations<br />

– 38.8 ~ 61.1% sum rate<br />

reduction<br />

– Interference dom<strong>in</strong>ated sum<br />

rate performance


Summary<br />

• The MU-<strong>MIMO</strong> with receive beamform<strong>in</strong>g comb<strong>in</strong>ed with<br />

multiuser is still an <strong>in</strong>terest<strong>in</strong>g alternative MU-<strong>MIMO</strong> subject to<br />

feedback channel errors.<br />

• Imperfect CST at the receiver (CRI-R) will cause serious sum-rate<br />

performance degradation and cause the sum-rate floor.<br />

– Sum rate capacity will no longer l<strong>in</strong>early <strong>in</strong>crease with<br />

SNR <strong>in</strong> decibel and be bounded.


Ma<strong>in</strong> Po<strong>in</strong>t <strong>in</strong> Story 3<br />

<strong>Multi</strong>-user broadcast <strong>MIMO</strong> systems DO NOT<br />

require transmit beamform<strong>in</strong>g precoders.<br />

“Schedul<strong>in</strong>g for <strong>Multi</strong>user <strong>MIMO</strong> Broadcast <strong>Systems</strong>: Transmit or Receive<br />

Beamform<strong>in</strong>g?” IEEE Trans. on Wireless Communications, Vol. 9, No. 9.<br />

pp. 2779~2791, Sep. 2010.


4. <strong>Multi</strong>-cell Network <strong>MIMO</strong><br />

systems<br />

Story 4: Channel assignment for multi-cell<br />

Network <strong>MIMO</strong>


What and Why Network <strong>MIMO</strong>?<br />

• Inter-cell <strong>in</strong>terference: In conventional cellular systems, the<br />

solution is us<strong>in</strong>g larger frequency reuse among cells<br />

– Poorer spectral efficiency!!<br />

Inter-stream <strong>in</strong>terference free!!<br />

<strong>MIMO</strong> broadcast system<br />

CSI exchange through high-speed backbaul<br />

Inter-cell <strong>in</strong>terference free!!<br />

Network <strong>MIMO</strong> system<br />

The concept is also used <strong>in</strong> 3GPP LTE-A (CoMP) and IEEE 802.16m (Co-<strong>MIMO</strong>)


key Issues for Network <strong>MIMO</strong><br />

2nd-tier neighbor<strong>in</strong>g cells<br />

1st-tier neighbor<strong>in</strong>g cells<br />

Conventional omni-cell approach<br />

IGI causes serious signal<br />

quality degradation<br />

14 dB degradation for 7-cell network <strong>MIMO</strong><br />

23 dB degradation for 3-cell network <strong>MIMO</strong>


Literature Survey<br />

FFR?<br />

Sector or<br />

omni-cell?<br />

Network<br />

<strong>MIMO</strong>?<br />

Note<br />

[Lei’07],<br />

[Fujii’08]<br />

[Chiu’08]<br />

[Shamai’01]<br />

[Karakayali’06]<br />

o sector x SINR improvement and soft<br />

handoff scheme<br />

x omni o The concept of multi-base<br />

station cooperation<br />

[Somekh’06]<br />

[J<strong>in</strong>g’07]<br />

x<br />

Wyner’s<br />

model<br />

o<br />

Performance analysis based<br />

on Wyner’s circular cell model<br />

[Zhang’09] x omni o BD-based network <strong>MIMO</strong><br />

performance<br />

[Boccardi’07]<br />

[Huang’09]<br />

x sector o Effect of network <strong>MIMO</strong><br />

+ 3, 6, 12 sector per cell<br />

Our work o sector o Design a 3-cell network <strong>MIMO</strong><br />

<strong>in</strong> FFR tri-sector cell


Basic Fractional Frequency Reuse (FFR) Concept


FFR <strong>in</strong> Tri-Sector <strong>Cell</strong>*<br />

(Regular Approach) <br />

Each cell has the same frequency partition!!<br />

*F. Khan, LTE for 4G Mobile Broadband: Air Interface Technologies and Performance, 1st ed. Cambridge University Press, 2009.


Proposed Solution<br />

Coord<strong>in</strong>ated group under<br />

certa<strong>in</strong> frequency band<br />

FFR: fractional frequency reuse


Regular versus Rearranged Frequency Partitions<br />

A cell can simultaneously forms 3 groups with neighbor<strong>in</strong>g 6 cells


Benefit of Rearranged Frequency Partition<br />

• Extra 8.5 dB SINR improvement!!<br />

(compared to traditional tri-sector FFR)<br />

10 dB improvement for sector<strong>in</strong>g FFR<br />

Extra 2.7 dB ga<strong>in</strong> under regular<br />

partition + 3-cell network <strong>MIMO</strong><br />

Extra 8.5 dB ga<strong>in</strong> under rearranged<br />

partition + 3-cell network <strong>MIMO</strong><br />

90% percentile


Proposed Architecture with 60 o Sector<strong>in</strong>g


Benefit of Proposed Architecture<br />

over Conventional Approach<br />

• A smaller coord<strong>in</strong>ation, 3-cell network <strong>MIMO</strong> architecture, can<br />

even outperform conventional 7-cell network <strong>MIMO</strong> system<br />

– 2 dB SINR improvement for 60 o sector<strong>in</strong>g<br />

– 3 dB SINR improvement for 120 o sector<strong>in</strong>g<br />

Traditional 7-cell approach<br />

Proposed 3-cell approach<br />

with rearranged partition


Benefit of <strong>Multi</strong>ple Antennas at the Base Station<br />

• Each sector BS equips with t transmit antennas to serve t<br />

users simultaneously.<br />

• 3-cell network coord<strong>in</strong>ation with proposed rearranged trisector<br />

frequency partition.<br />

– Total 3t transmit antennas serve 3t user term<strong>in</strong>als<br />

Almost l<strong>in</strong>ear sum rate improvement<br />

as number of antennas t <strong>in</strong>creases<br />

Proposed network <strong>MIMO</strong> architecture<br />

achieves extra ga<strong>in</strong> over conventional<br />

approach


Summary<br />

• By exploit<strong>in</strong>g geographic cell distribution and frequency<br />

partition, a 3-cell network <strong>MIMO</strong> can outperform<br />

conventional 7-cell network <strong>MIMO</strong> with omni-directional cell<br />

• This k<strong>in</strong>d of 3-cell coord<strong>in</strong>ated network <strong>MIMO</strong> architectures<br />

is particularly useful<br />

– The default number of neighbor<strong>in</strong>g cells <strong>in</strong> the Co-<strong>MIMO</strong><br />

transmission is three for IEEE 802.16m<br />

– In LTE-A RAN 1 meet<strong>in</strong>g, the number of coord<strong>in</strong>ation <strong>in</strong><br />

CoMP is also three<br />

CoMP: coord<strong>in</strong>ated multi-po<strong>in</strong>t<br />

Co<strong>MIMO</strong>: collaborative multiple-<strong>in</strong>put multiple-output


Ma<strong>in</strong> Po<strong>in</strong>t <strong>in</strong> Story 4<br />

Network <strong>MIMO</strong> systems only need 3 cell site<br />

to cooperate<br />

• Li-Chun Wang and Chu-Jung Yeh, “3-<strong>Cell</strong> network <strong>MIMO</strong> architectures<br />

with sectorization and fractional frequency reuse,” IEEE Journal <strong>in</strong><br />

Selected Area <strong>in</strong> Communications, Vol. 29, No. 6, pp. 1185~1199,<br />

June, 2011<br />

• Li-Chun Wang and Chu-Jung Yeh, “Antenna architectures for network<br />

<strong>MIMO</strong>,” Cooperative <strong>Cell</strong>ular Wireless Networks,” Cambridge University<br />

Press, ISBN-13: 9780521767125.<br />

• IEEE C802.16m-09/2280 “Frequency Plann<strong>in</strong>g for Inter-<strong>Cell</strong><br />

Interference Reduction <strong>in</strong> 3-<strong>Cell</strong> Collaborative <strong>MIMO</strong> <strong>Systems</strong>”


5. Updates of Network <strong>MIMO</strong><br />

Techniques <strong>in</strong> 3GPP LTE-A<br />

Updates


3GPP LTE/HSPA Evolution <br />

HSPA Series Evolution<br />

(CDMA-based)<br />

• DL: 2x2 <strong>MIMO</strong><br />

• DL: 2x2 <strong>MIMO</strong>+64QAM<br />

or DC+64QAM<br />

• DL: DC+64QAM+<strong>MIMO</strong><br />

• UL: DC<br />

• DL: 4-Carrier<br />

LTE Series Evolution<br />

(OFDM-based)<br />

Note: Dates refer to the first completed (full) specifications<br />

81<br />

• DL: 4x4 <strong>MIMO</strong> <strong>in</strong> 20MHz<br />

3GPP Concept for<br />

IMT-A (4G)


LTE-A Key Technologies<br />

• Carrier Aggregation<br />

– Aggregate bandwidth up to 100MHz<br />

• Downl<strong>in</strong>k transmission scheme<br />

– Improvements to LTE by us<strong>in</strong>g 8x8 <strong>MIMO</strong><br />

– Data rates of 100Mb/s with high mobility and 1Gb/s<br />

with low mobility<br />

• Upl<strong>in</strong>k transmission scheme<br />

– Improvements to LTE by us<strong>in</strong>g 4x4 <strong>MIMO</strong><br />

– Data rates up to 500Mb/s<br />

• Relay Functionality<br />

– Improv<strong>in</strong>g cell edge coverage<br />

– More efficient coverage <strong>in</strong> rural areas<br />

82


Advanced <strong>MIMO</strong> Techniques <strong>in</strong> DL<br />

• Extension up to 8-stream transmission<br />

– Increased from 4 streams <strong>in</strong> Rel-8/9<br />

– Satisfy peak SE requirement (i.e., 30 bps/<br />

Hz)<br />

• Support for enhanced MU-<strong>MIMO</strong><br />

– Not more than 4 UEs are co-scheduled<br />

– Not more than 2 layers per UE<br />

83


Advanced <strong>MIMO</strong> Techniques <strong>in</strong> UL<br />

• Introduction of SU-<strong>MIMO</strong> up to 4-stream<br />

transmission<br />

– Rel. 8 LTE does not support SU-<strong>MIMO</strong><br />

– Satisfy peak SE requirement (i.e., 15 bps/<br />

Hz)<br />

• Introduction of UL transmit diversity for<br />

PUCCH<br />

– Improved signal<strong>in</strong>g robustness and celledge<br />

performance<br />

84


Enhanced ICIC<br />

• Enhanced ICIC for non-CA based deployments<br />

of heterogeneous networks for LTE<br />

– WI proposed by CMCC <strong>in</strong> RAN#47 (Mar. 2010)<br />

– Identify and evaluate non-CA based ICIC strategies<br />

of heterogeneous network deployments<br />

– HetNet use cases are priorities as follows:<br />

» Indoor HeNB clusters<br />

» Outdoor Hotzone cells<br />

» Indoor Hotzone scenarios<br />

85


R10 LTE-A Performance Improvement<br />

• Rel-8 LTE vs. IMT-A requirements<br />

– DL: Rel-8 LTE fulfills IMT-A requirements<br />

– UL: Need to double from Rel-8 to satisfy<br />

IMT-A requirement<br />

FDD Rel-8 LTE (1) Rel-10LTE-A (2) IMT-Advanced<br />

Peak spectrum efficiency<br />

[bps/Hz]<br />

DL 16.3 30.6 15<br />

UL 4.3 16.8 6.75<br />

TDD Rel-8 LTE (1) Rel-10LTE-A (2) IMT-Advanced<br />

Peak spectrum efficiency<br />

[bps/Hz]<br />

DL 16 30 15<br />

UL 4 16.1 6.75<br />

Note:<br />

(1)<br />

4x4 <strong>MIMO</strong> <strong>in</strong> DL, SIMO <strong>in</strong> UL<br />

(2)<br />

8x8 <strong>MIMO</strong> <strong>in</strong> DL, 4x4 <strong>MIMO</strong> <strong>in</strong> UL <br />

Source: 3GPP TR 36.912<br />

86


R11 LTE-A New Technical Issues<br />

• CoMP (Coord<strong>in</strong>ated <strong>Multi</strong>po<strong>in</strong>t Transmission and Reception)<br />

• MTC (Mach<strong>in</strong>e-Type Communications )<br />

• Network Energy Sav<strong>in</strong>g<br />

• MODAI (Mobile Data Impact)<br />

87


CoMP (Coord<strong>in</strong>ated <strong>Multi</strong>po<strong>in</strong>t Transmission and Reception)<br />

• New study Item <strong>in</strong> Rel-10<br />

– Approved <strong>in</strong> RAN#47 (Mar. 2010)<br />

• DL CoMP Transmission Scheems<br />

– Jo<strong>in</strong>t Process<strong>in</strong>g<br />

» Jo<strong>in</strong>t transmission/dynamic cell selection<br />

– Coord<strong>in</strong>ated Schedul<strong>in</strong>g/Beamform<strong>in</strong>g<br />

• Better cell edge performance<br />

– Less co-channel <strong>in</strong>terference / more signal<br />

88


Jo<strong>in</strong>t Process<strong>in</strong>g (JP)<br />

The coord<strong>in</strong>ated cells will serve the user together. The base<br />

station will exchange the channel state <strong>in</strong>formation (CSI)<br />

and data to each other.<br />

CSI and data exchange between each base station<br />

Coord<strong>in</strong>ated transmission to the<br />

user


Coord<strong>in</strong>ated Beamform<strong>in</strong>g (CB)<br />

If there are two users close to each other but <strong>in</strong> different cells.<br />

By sett<strong>in</strong>g the appropriate beam<strong>in</strong>g weight, we can decrease the<br />

<strong>in</strong>terference from different base stations. <br />

Only CSI exchange between each base station<br />

Only one base station will transmit to the user


Coord<strong>in</strong>ated Schedul<strong>in</strong>g (CS)<br />

Use the coord<strong>in</strong>ated schedul<strong>in</strong>g to decide which user will be<br />

serve. Even without <strong>in</strong>terference null<strong>in</strong>g, we can avoid the<br />

<strong>in</strong>terference from the other base stations.<br />

Only CSI exchange between each base station<br />

Only one base station will transmit to the appropriate user


Mulit-<strong>Cell</strong> Co-operations<br />

Jo<strong>in</strong>t Process<strong>in</strong>g<br />

for <strong>User</strong> Data<br />

Shared CSI <strong>in</strong><br />

<strong>Multi</strong>-Sites<br />

Jo<strong>in</strong>t<br />

DataTransmission<br />

Interference<br />

Null<strong>in</strong>g<br />

JP CB CS<br />

☺ X X<br />

☺ ☺ ☺<br />

group 1 1<br />

☺ ☺ X<br />

Beamform<strong>in</strong>g ☺ ☺ ☺<br />

JP : Jo<strong>in</strong>t Process<strong>in</strong>g<br />

CB : Coord<strong>in</strong>ated<br />

Beamform<strong>in</strong>g<br />

CS : Coord<strong>in</strong>ated Schedul<strong>in</strong>g


4 Scenarios for CoMP <strong>in</strong> 3GPP LTE-A <br />

• Accord<strong>in</strong>g to R1-110564 <strong>in</strong> 3GPP, CoMP (coord<strong>in</strong>ated<br />

multipo<strong>in</strong>t transmission) techniques are applied <strong>in</strong> the<br />

follow<strong>in</strong>g 4 scenario<br />

Scenario 1 : Homogeneous network with <strong>in</strong>tra-site<br />

Scenario 3 : Heterogeneous network with low power<br />

RRHs, which create cells with their own cell ID.<br />

Scenario 2 : Homogeneous network with high Tx power<br />

RRHs<br />

Scenario 4 : Heterogeneous network with low power<br />

RRHs, which create cells with the same cell ID as the macro.


NCTU has built a LTE-A System-Level Simulator<br />

• In order to propose new techniques for LTE-A systems, we must<br />

establish the LTE-A system-level simulator and calibrate the<br />

performance metrics <strong>in</strong> 3GPP TR36.814.<br />

Step(1a) : wideband<br />

SINR<br />

Path-loss, Shadow<strong>in</strong>g, Antenna Pattern<br />

Step(1c) : DL SINR and cell<br />

spectral efficiency<br />

Spatial Channel Model, OFDMA, MRC Antenna<br />

Comb<strong>in</strong><strong>in</strong>g, Adaptive MCS , HARQ<br />

SU/MU-<strong>MIMO</strong><br />

Codebook-based Precoder, MMSE receiver, Rank<br />

Adaptation, SU/MU Mode Switch<strong>in</strong>g<br />

Hierarchical Network<br />

<strong>MIMO</strong><br />

RRH Selection, Cooperation/Non-cooperation<br />

mode switch<strong>in</strong>g


<strong>MIMO</strong> Physical Layer Simulation Flow Chart


NCTU LTE-A System-Level Simulator Interface


Hierarchical Base Station Cooperation with<br />

S<strong>in</strong>gle <strong>Cell</strong> ID (HBSC-S)<br />

• If the RRH nodes share the same cell ID with the correspond<strong>in</strong>g<br />

macro-BS, the RRH nodes can be regarded as the distributed<br />

antennas of the macro-BS. We call the system as Hierarchical<br />

Base Station Cooperation with the S<strong>in</strong>gle cell ID (HBSC-S).


Hierarchical Base Station Cooperation with<br />

<strong>Multi</strong>ple <strong>Cell</strong> IDs (HBSC-M)<br />

• We apply the cooperation technique to mitigate the <strong>in</strong>tra-cell<br />

<strong>in</strong>terference.<br />

• We call the system as Hierarchical Base Station Cooperation with<br />

<strong>Multi</strong>ple cell IDs (HBSC-M).


Wideband SINR for Calibration Step 1a<br />

Our calibration result is consistent with those <strong>in</strong><br />

3GPP.


Spectral Efficiency for Calibration Step 1c<br />

Our calibration result is consistent with those <strong>in</strong><br />

3GPP.


Spectral Efficiency for <strong>MIMO</strong> <strong>Systems</strong><br />

• All our simulation results are consistent with those <strong>in</strong><br />

3GPP.<br />

Spectral<br />

Efficiency<br />

2x2 SU-<strong>MIMO</strong><br />

<strong>Cell</strong>-Average<br />

2x2 SU-<strong>MIMO</strong><br />

5% <strong>Cell</strong>-Edge<br />

4x2 SU-<strong>MIMO</strong><br />

<strong>Cell</strong> Average<br />

4x2 SU-<strong>MIMO</strong><br />

5% <strong>Cell</strong>-Edge<br />

2x2 MU-<strong>MIMO</strong><br />

<strong>Cell</strong>-Average<br />

2x2 MU-<strong>MIMO</strong><br />

5% <strong>Cell</strong>-Edge<br />

M<strong>in</strong>-value of<br />

others <strong>in</strong> 3GPP<br />

(bits/s/Hz)<br />

Our Work<br />

(bits/s/Hz)<br />

Max-value of<br />

others <strong>in</strong> 3GPP<br />

(bits/s/Hz)<br />

2.14 2.37 2.47<br />

0.072 0.079 0.100<br />

2.34 2.52 2.66<br />

0.085 0.089 0.110<br />

2.56 2.62 2.77<br />

0.070 0.086 0.110


Spectral Efficiency for HBSC-M <strong>Systems</strong> <strong>in</strong> <strong>Multi</strong>-<strong>Cell</strong><br />

Case<br />

• HBSC-M systems outperform both the HBSC-S systems and the SU-<br />

<strong>MIMO</strong> system <strong>in</strong> the spectral efficiency.<br />

• The best position for each RRH node is 0.6R~0.7R.<br />

+12%<br />

+ 20%


Energy Efficiency for HBSC <strong>Systems</strong><br />

• When consider<strong>in</strong>g the energy efficiency (bits/<br />

Joule), the HBSC-S system with 1 RRH selected<br />

outperforms the HBSC-M system.<br />

+ 9%


Ma<strong>in</strong> Po<strong>in</strong>t <strong>in</strong> Story 5<br />

LTE-A system-level simulator support<strong>in</strong>g<br />

hierarchical network <strong>MIMO</strong> is important.<br />

3GPP R1-111528, “Simulation Evaluation for CoMP Scenarios 3 and 4”,<br />

May 2011<br />

“Performance Calibration and Simulation Methodology for<br />

3GPP LTE-A <strong>Systems</strong>.” IEEE VTS APWCS, Aug. 2011<br />

2011/7


Conclusions<br />

• We have discussed multiuser <strong>MIMO</strong> communications<br />

systems from the perspective of radio resource arrangement<br />

and place emphasis on user schedul<strong>in</strong>g and channel<br />

assignment correctly.<br />

• Ma<strong>in</strong> po<strong>in</strong>t: If we utilize the degree of freedom <strong>in</strong> the user<br />

doma<strong>in</strong> cleverly, we can deliver many advantages to MU-<br />

<strong>MIMO</strong> and network <strong>MIMO</strong>:<br />

– enhanc<strong>in</strong>g coverage and capacity simultaneously;<br />

– Improv<strong>in</strong>g robustness to channel variations;<br />

– Reduc<strong>in</strong>g the necessity of large number of cooperative<br />

cell site.


Future Research Directions<br />

• Channel state feedback for downl<strong>in</strong>k network <strong>MIMO</strong><br />

• Asynchronous upl<strong>in</strong>k network <strong>MIMO</strong><br />

• Hierarchical network <strong>MIMO</strong>


F<strong>in</strong>al remark<br />

Com<strong>in</strong>g together is a beg<strong>in</strong>n<strong>in</strong>g.<br />

Keep<strong>in</strong>g together is progress.<br />

Work<strong>in</strong>g together is success.<br />

~ Henry Ford


References<br />

• C. J. Chen and Li-Chun Wang, "Enhanc<strong>in</strong>g Coverage and Capacity for<br />

<strong>Multi</strong>user <strong>MIMO</strong> <strong>Systems</strong> by Utiliz<strong>in</strong>g Schedul<strong>in</strong>g, ” IEEE Trans. on Wireless<br />

Communications, Vol. 5, No. 5, pp. 1148-1157, May, 2006.<br />

• C. J. Chen and Li-Chun Wang, “A Unified Capacity Analysis for Wireless<br />

<strong>Systems</strong> with Jo<strong>in</strong>t <strong>Multi</strong>user Schedul<strong>in</strong>g and Antenna Diversity <strong>in</strong><br />

Nakagami Fad<strong>in</strong>g Channels,” IEEE Trans. on Communications, vol. 54, No.<br />

3, pp. 469~478, Mar. 2006.<br />

• Li-Chun Wang and Chu-Jung Yeh, “Schedul<strong>in</strong>g for <strong>Multi</strong>user <strong>MIMO</strong><br />

Broadcast <strong>Systems</strong>: Transmit or Receive Beamform<strong>in</strong>g?” IEEE Trans. on<br />

Wireless Communications, Vol. 9, No. 9. pp. 2779~2791, Sep. 2010.<br />

• Li-Chun Wang and Chu-Jung Yeh, “3-<strong>Cell</strong> network <strong>MIMO</strong> architectures<br />

with sectorization and fractional frequency reuse,” IEEE Journal <strong>in</strong><br />

Selected Area <strong>in</strong> Communications, Vol. 29, No. 6, pp. 1185~1199, June,<br />

2011<br />

• Li-Chun Wang and Chu-Jung Yeh, “Antenna architectures for network<br />

<strong>MIMO</strong>,” Cooperative <strong>Cell</strong>ular Wireless Networks,” Cambridge University<br />

Press, ISBN-13: 9780521767125.<br />

• IEEE C802.16m-09/2280 “Frequency Plann<strong>in</strong>g for Inter-<strong>Cell</strong> Interference<br />

Reduction <strong>in</strong> 3-<strong>Cell</strong> Collaborative <strong>MIMO</strong> <strong>Systems</strong>”


References<br />

• Y. J. Liu, T. T. Chiang and Li-Chun Wang “Performance Calibration and<br />

Simulation Methodology for 3GPP LTE-A <strong>Systems</strong>.” IEEE VTS APWCS, Aug.<br />

2011<br />

• 3GPP R1-111528, “Simulation Evaluation for CoMP Scenarios 3 and 4”,May<br />

2011<br />

• -------------------------------------------------------------------------------------


Reference<br />

[1] TR 36.814 : 3rd Generation Partnership Project ; Technical Specification<br />

Group Radio Access Network; Evolved Universal Terrestrial Radio Access (E-<br />

UTRA); Further advancements for E-UTRA physical layer aspects (Release 9)<br />

[2] TR 36.819 : 3rd Generation Partnership Project;Technical Specification<br />

Group Radio Access Network; Coord<strong>in</strong>ated multi-po<strong>in</strong>t operation for LTE<br />

physical layer aspects(Release 11)<br />

[3] M.2135 Report ITU-R M.2135-1(12/2009) Guidel<strong>in</strong>es for evaluation of radio<br />

<strong>in</strong>terface technologies for IMT-Advanced M Series Mobile,<br />

radiodeterm<strong>in</strong>ation, amateur and related satellites services<br />

[4] S. Shamai and B. M. Zaidel, “Enhanc<strong>in</strong>g the cellular downl<strong>in</strong>k capacity via<br />

co-process<strong>in</strong>g at the transmitt<strong>in</strong>g end,” IEEE Vehicular Technology<br />

Conference Spr<strong>in</strong>g, vol. 3, pp. 1745 – 1749, May 2001.<br />

[5] H. Zhang and H. Dai, “Cochannel <strong>in</strong>terference mitigation and<br />

cooperative process<strong>in</strong>g <strong>in</strong> downl<strong>in</strong>k multicell multiuser <strong>MIMO</strong> networks,”<br />

EURASIP Journal on Wireless Communications and Network<strong>in</strong>g, vol. 2004, no.<br />

2, pp. 222–235, 2004.


Reference<br />

[6] G. J. Fosch<strong>in</strong>i, K. Karakayali, and R. A. Valenzuela, “Coord<strong>in</strong>at<strong>in</strong>g multiple<br />

antenna cellular networks to achieve enormous spectral efficiency,” IEE<br />

Proceed<strong>in</strong>gs-Communications, vol. 153, no. 4, pp. 548 – 555, Aug. 2006.<br />

[7] S. J<strong>in</strong>g, D. N. C. Tse, J. B. Soriaga, J. Hou, J. E. Smee, and R. Padovani,<br />

“<strong>Multi</strong>cell downl<strong>in</strong>k capacity with coord<strong>in</strong>ated process<strong>in</strong>g,” EURASIP<br />

Journal on Wireless Communications and Network<strong>in</strong>g, 2008.<br />

[8] H. Huang, M. Trivellato, A. Hott<strong>in</strong>en, M. Shafi, P. Smith, and R. Valenzuela,<br />

“Increas<strong>in</strong>g downl<strong>in</strong>k cellular throughput with limited network <strong>MIMO</strong><br />

coord<strong>in</strong>ation,” IEEE Transactions on Wireless Communications, vol. 8, no.<br />

6, Jun. 2009.<br />

[9] F. Richter, A. J. Fehske, and G. Fettweis, “Energy efficiency aspects of<br />

base station deployment strategies for cellular networks,” IEEE Vehicular<br />

Technology Conference Fall, Sep. 2009.<br />

[10] A. J. Fehske, P. Marsch, and G. P. Fettweis, “Bit per Joule efficiency of<br />

cooperat<strong>in</strong>g base stations <strong>in</strong> cellular networks,” IEEE GLOBECOM<br />

Workshops, Dec. 2010.


Reference<br />

[11] J. Salo, G. Del Galdo, J. Salmi, P. Kyosti, M. Milojevic, D. Laselva, and C.<br />

Schneider. (2005, Jan.) MATLAB implementation of the 3GPP Spatial<br />

Channel Model (3GPP TR 25.996). [Onl<strong>in</strong>e]. Available: http://www.tkk.fi/<br />

Units/Radio/scm<br />

[12] 3GPP, R1-110867, “CoMP with lower Tx power RRH <strong>in</strong> heterogeneous<br />

network,” NTT DOCOMO, Feb. 21-25, 2011.<br />

[13] L. Thiele, T. Wirth, M. Schellmann, Y. Hadisusanto, and V. Jungnickel, “MU-<br />

<strong>MIMO</strong> with localized downl<strong>in</strong>k base station cooperation and downtilted<br />

antennas,” IEEE International Conference on Communications<br />

Workshops, pp. 1–5, 2009.<br />

[14] N. J<strong>in</strong>dal, “Antenna comb<strong>in</strong><strong>in</strong>g for the <strong>MIMO</strong> downl<strong>in</strong>k channel,” IEEE<br />

Transactions on Wireless Communications, vol. 7, no. 10, pp. 3834–3844,<br />

Oct. 2008.<br />

[15] S. Fang, G. Wu, Y. Xiao, and S. Q. Li, “<strong>Multi</strong>-user <strong>MIMO</strong> l<strong>in</strong>ear precod<strong>in</strong>g<br />

with grassmannian codebook,” International Conference on<br />

Communications and Mobile Comput<strong>in</strong>g, vol. 1, pp. 250–255, 2009


Reference<br />

[16]. M. Sawahashi, Y. Kishiyama, A. Morimoto, D. Nishikawa, and M. Tanno,<br />

“Coord<strong>in</strong>ated multipo<strong>in</strong>t transmission/ reception techniques for LTE-<br />

Advanced,” IEEE Wireless Commun. Mag., pp. 26-34, June 2010.<br />

[17]. S. A. Jafar, “Interference Alignment: A New Look at Signal Dimensions <strong>in</strong><br />

a Communication Network,” Foundations and Trends <strong>in</strong> Communications<br />

and Information Theory, 2011.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!