Fractional Base Station Cooperation Cellular · PDF fileFractional Base Station Cooperation...

Fractional Base Station Cooperation

Cellular Network

Naoki Kusashima

Tokyo Institute of Technology,

Department of Electrical and Electronic Engineering,

Araki-Sakaguchi Laboratories.

Contents• Background

– Cell-edge problem

– Conventional solution for the cell-edge problem

– Base station cooperation

• Base station cooperation block diagonalization

multi-user MIMO

– Combination of multi-user

– Cooperative region

• Fractional base station cooperation cellular network

– Fractional Base Station Cooperation

– Dynamic clustering

• Numerical simulation

• Conclusion

2

Cell-edge problem

• If user at the cell-edge location

– Low SNR

– Co-channel interference from adjacent BSs

– High antenna correlation

Low user throughput

Signal

Interference1st path

Co-channel interference High antenna correlation

2nd path

Cell-edge

problem

3

Conventional solution for cell-edge problem

• Conventional solution

– Frequency reuse

• Each BSs uses different frequency channel from surrounding

cells.

It causes degrading of system throughput.

– Interference cancellation using terminal adaptive array

• cancels interference by terminals.

It requires many antennas to perform multiplexing and

cancellation at the same time.Interference cancellation

using terminal adaptive arrayFrequency reuse

4

Base station cooperation• Base Station Cooperation (BSC)

– BSC transmits multiple streams to cooperate with adjacent BSs.

– There are no problems of nor inter-cell interference nor high antenna

correlation.

BSC single-user (BSC-SU)

• For single user

BSC-SU wastes many resources to a user.

BSC multi-user (BSC-MU)

• For multiple users

BSC-MU uses resource more efficient than BSC-SU.

Base station cooperation

Single-User

Base station cooperation

Multi-User

5

Base station cooperation block

diagonalization multi-user MIMO

2nd BS

111 dg

112 dDg 221 dDg

222 dg

1d 2d2nd user1st BS 1st user

• Transmission model of (KBS,M)×(KMS,N) BSC-MU-MIMO

：pathloss between jth BS and ith user：distance between ith BS and ith user ijgid

D

• Receive signal of ith user

iiKiKiiiii nxHxHxHy BSBS2211

ijH : channel matrix

ijx : transmit signal vectorin : noise vector

iy : receive signal vector

iiiii nsVQH ~

iQ : precoding matrix of

block diagonalization

is : transmit data signal vector6

iV~

: precoding matrix of

SVD-MIMO

Base station cooperation block

diagonalization multi-user MIMO

• Block diagonalization

MSBSBSBSMSMS

BSBS

MS

11

11

11111

~~

~~

KKKKKK

KK

K n

n

s

s

VQHVQH

VQHVQH

y

y

MSBSBSMS

1111

~~

~~

KKKK n

n

s

s

VHO

OVH

Channel matrix is block diagonalized

H||

\\svd iii QVOUH i

TT

K

T

i

T

i

T

\111 MS,...,,,..., HHHHH

• Precoding matrix Q

ij

ijiij

O

HQH

~

7

Base station cooperation block

diagonalization multi-user MIMO

iid

ijH ：fading matrix

• We decompose the channel matrix into fading matrix

and pathloss matrix

IG ijij g ii

i

i

iiiii GHGO

OGHHHHH

iid

2

1iid

2

iid

121

iid1

1

1ijij

Fij

iji HGG

HQ OHH

iidiid

ijij

• Block diagonalization precoding matrix of BSC

OQH iji

• Receive signal of ith user

iiiiiiiiii nsVHnsVQHy ~~~

iid

21

21122211iid1iid

1

~1~i

ijij

ijijii

Fij

iiigg

ggggHHGGH

GQHH

iid~iH ：The equivalent fading matrix of ith user

8

Base station cooperation block

diagonalization multi-user MIMO

• Power normalization factor

– Ω regulates the transmit power

2

1

2 1logk

ikiC

• Capacity of ith user

21

2221

21

1211

11

21

2221

22

1211

12

1

1

dd

gg

g

gg

g

dd

gg

g

gg

g

• Average receive SNR of ith user

k ：kth eigenvalue of equivalent channel

PEP j

H

jj xx P

：maximum BS transmit power

：power normalization factor

2

21

21122211

2

2~~

P

gg

gggg

M

E

ijij

iii

i

sVH

• Transmit power from jth BS

9

jKjj MS,...,1 xxx

：transmit signal of jth BS

d1 [m]

d2 [

m]

0 100 200 300 400 5000

100

200

300

400

500

0.7

0.75

0.8

0.85

0.9

0.95

1

Power normalization factor Ω

BS2MS2MS1BS1

1d 2d

• d1=d2

Ωmax=1

• d1>>d2, d1<<d2

Ωmin=2/3

Co-schedulingThe BSs select users with the same SINR

Ω

10

0 100 200 300 400 5000

5

10

15

20

25

30

35

40

Distance from BS to user [m]

Av

era

ge u

ser1

cap

acit

y [

bp

s/H

z]

2x2 SC-MIMO Ideal

2x2 SC-MIMO

(2,2)x(2,2) BSC-MUco-schedule

(2,2)x(2,2) BSC-MUuser2 at cell-edge

Capacity of (2,2)x(2,2) BSC-MU-MIMO Average capacity of 1st user at different distance

• Co-scheduling

BSC-MU is almost the

same capacity as 2x2

SC-MIMO ideal

• User2 at cell-edge

Capacity is decreased

about 3bps where

distance is from 100m

to 300m

d1 [m]

d2 [

m]

0 100 200 300 400 5000

100

200

300

400

500

0.7

0.75

0.8

0.85

0.9

0.95

1

Co-scheduling

User2 at cell-edge

Co-scheduling is the optimal

user selection

Ω

11

Cooperative region

• Cell-inner

– Non-cooperative region

• Single-cell MIMO is

efficient at cell-inner

• Cell-edge

– Cooperative region

• BSC MIMO is efficient

at cell-edge

MC

MU

MC CR

SC

SU

SC CR

• Spectral efficiency

: bandwidth inefficiencies

considering overheadComparison of spectral efficiency

• Analysis considering overhead is

necessary– Overhead : the cost of resource for transmission

(ex. Reference signal, guard band, cyclic prefix)

12

Fractional Base Station Cooperation

• Single-cell MIMO (conventional cellular system)

– Cell-inner : efficient

– Cell-edge : not efficient

• Base station cooperation MIMO

– Cell-inner : not efficient

– Cell-edge : efficient

Fractional Base Station Cooperation (FBSC)

– Cell-inner：uses single-cell MIMO

– Cell-edge：uses BSC MIMO

FBSC is performed to achieve gains both at the cell-inner and cell-edge.13

Cooperation clustering• Clustering

– Static clustering

• Cooperation set is fixed

By-product cell-edge is created.

– Dynamic clustering

• makes cooperation set adaptively.

Cooperation is efficient at all cell-edges.

Static DynamicBy-product

cell-edge

14

Dynamic clustering

15

: master cell

: slave cell

Fractional base

station cooperation

Dynamic clustering

Fractional base station

cooperation cellular

network

Physical

layer

Fractional base station cooperation

cellular network

Network

layer

Distributed

controller

Cooperative

region

Non-cooperative

region 16

Simulation scenario• Comparison transmission

– Single-cell SISO

– Single-cell MIMO

• 2x2 SVD-MIMO

– Multi-cell static clustering

• BSC is performed at all locations

• Static clustering

– Fractional Base Station

Cooperation Cellular Network

(FBSC-CN)

17

Cooperative region

Non-cooperative region

Cooperative region

Cooperative region at 19-cell scenario

Simulation scenario

Parameters Values

Number of cell 19 cells

Number of user 10 users / cell

Number of cooperation set 3 cells

Number of antenna

( BS, user )1, 1 (SISO) or 2, 2 (MIMO)

Channel model i.i.d. Rayleigh

Pathloss model34.5 + 35log10(d[m]) [dB]

(3GPP TR 25.996 : Urban Macro)

Transmit power 40[dBm]

Noise level -100[dBm]

Site to site distance 1000 m

Scheduler Round-robin & co-scheduling

Single-user MIMO scheme SVD-MIMO

Multi-user MIMO scheme Generalized BD [*]

Simulation parameter

18* V. Stankovic, M. Haardt, “Generalized design of multi-user MIMO precoding matrices,” IEEE Trans. Wireless

Commun., vol.7, no.3, pp.953-961, Mar. 2008.

100 200 300 400 5000

0.2

0.4

0.6

0.8

1

Distance from BS to user [m]

Use

r sp

ectr

al

eff

icie

ncy

[b

ps/

Hz/u

ser]

Single-cell SISO

Single-cell MIMO

Frequency reuse

Interference cancellation

Multi-cell static clustering

FBSC-CN

100 200 300 400 5000

0.2

0.4

0.6

0.8

1

Distance from BS to user [m]

Use

r sp

ectr

al

eff

icie

ncy

[b

ps/

Hz/u

ser]

Single-cell SISO

Single-cell MIMO

Frequency reuse

Interference cancellation

Multi-cell static clustering

FBSC-CN

100 200 300 400 5000

0.2

0.4

0.6

0.8

1

Distance from BS to user [m]

Use

r sp

ectr

al

eff

icie

ncy

[b

ps/

Hz/u

ser]

Single-cell SISO

Single-cell MIMO

Frequency reuse

Interference cancellation

Multi-cell static clustering

FBSC-CN

Simulation result

19

Fractional BSC

• User capacity at

the cell-inner is as

high as Single-

cell MIMO

• User capacity at

the cell-edge is

almost the same

with BS

cooperation

User capacity at the distance locations

0 0.2 0.4 0.6 0.8 1 0

0.05

0.2

0.4

0.6

0.8

1

User spectral efficiency [bps/Hz]

CD

F

Single-cell SISO

Single-cell MIMO

Multi-cell static clustering

FBSC-CN

0 0.2 0.4 0.6 0.8 1 0

0.05

0.2

0.4

0.6

0.8

1

User spectral efficiency [bps/Hz]

CD

F

Single-cell SISO

Single-cell MIMO

Multi-cell static clustering

FBSC-CN

Simulation result

20

• FBSC is higher

capacity than

other scheme at

all probability

CDF of instantaneous capacity at cell-edge user

Simulation result

21

0

1

2

3

4

Av

era

ge c

ell

spectr

al

eff

icie

ncy

[bp

s/H

z]

0

0.05

0.1

0.15

0.2

Cell

-ed

ge u

ser

spectr

al

eff

icie

ncy

[bp

s/H

z/u

ser]

w/o shadow fading

w/ shadow fading

w/o shadow fading

w/ shadow fading

Single-cell

SISO

Single-cell

MIMO

Single-cell

SISO

Single-cell

MIMO

Multi-cell

Static clustering

FBSC-CN

Multi-cell

Static clustering

FBSC-CN

• The average cell spectral efficiency of FBSC is slightly improved than

single-cell MIMO

• The cell-edge user spectral efficiency of FBSC is 2.4 times as high as

that of single-cell MIMO

Average cell spectral efficiency and cell-edge spectral efficiency

(shadow fading standard deviation = 8[dB])

Conclusion

• BSC solves the cell-edge problem.

• Fractional BSC is proposed.

– FBSC performs to achieve gains both at the cell-inner

and cell-edge.

• FBSC cellular network is proposed.

– In FBSC-CN, cooperation sets are constructed

dynamically.

• Numerical simulation shows that FBSC

performs efficiently both at cell-inner and cell-

edge.

22

Thank you for your attention.

23