Coverage in Heterogeneous Networks · Femtocells • Femtocells are low-power wireless access...

Coverage in Heterogeneous Coverage in Heterogeneous Networks

Xiaoli ChuKing’s College London

UC4G Beijing Workshop August 2010

Outline• Introduction

▫ Heterogeneous networksHeterogeneous networks▫ Challenges

• Coverage in heterogeneous networks▫ System model▫ Outage probability analysis▫ Maximum density of co-channel femtocell transmissionsy▫ Femtocell transmit power

• Simulation resultsA l ti l lt ifi d b i l ti▫ Analytical results verified by simulations

▫ Low attenuation vs. high attenuation environments• Conclusions• Future work

UC4G Beijing Workshop August 2010 - 2 -

Introduction


Why Heterogeneous?Why Heterogeneous?• Needs for substantial improvements in cellular capacity

and coverageand coverage▫ Radio link improvement alone cannot meet increasing demands▫ Traffic demand and channel condition vary with time and location

• Improve spectral efficiency/area/cost▫ Need to increase cell density cost-effectively▫ Network topology provides gains beyond radio technologyp gy p g y gy

• Heterogeneous networks▫ Deploy macro-, pico-, femto-cells, and relays in the same spectrum

I d it th h b tt ti l▫ Improve coverage and capacity through better spatial reuse▫ Address hot-spot needs and coverage holes▫ Lower traffic load on macrocells


DeploymentDeployment• Initial coverage with macro base stations (MBS)• Add pico, femto and relay stations for capacity increase, hot spots,

indoor coverage, etc.• Pico, femto and relay stations require lower power, offer flexible

site acquisition, and add little or no backhaul expense.

Source: Intel Labs


Heterogeneous NetworksHeterogeneous Networks• Macrocells

▫ Operator-deployed BSs use dedicated backhaulp p y▫ Open to public access▫ PTx ~ 43 dBm, G ~ 12-15 dBi

• PicocellsPicocells ▫ Operator-deployed BSs use dedicated backhaul▫ Open to public access▫ PT ~ 23-30 dBm G ~ 0-5 dBiPTx 23 30 dBm, G 0 5 dBi

• Femtocells▫ User-deployed BSs use user’s broadband connection as backhaul▫ Open access closed access or a hybrid access policy▫ Open access, closed access, or a hybrid access policy▫ PTx ≤ 23 dBm, G ~ 0-2 dBi

• RelaysO t d l d BS th i li k t MBS b kh l▫ Operator-deployed BSs use over-the-air link to MBS as backhaul

▫ PTx ~ 23-30 dBm, G ~ 0-5 dBi


HN CharacteristicsHN Characteristics• Coexistence of potentially different access technologies• Different cell scales: macro > micro > pico > femto• BSs owned by operators, enterprises and consumers

Wi d i l b kh l ith t d b t ff t Q S• Wired or wireless backhaul with guaranteed or best-effort QoS• Femtocells potentially offer coverage only to subscribed UEs

Source: Qualcomm


ChallengesChallenges• Maintain uniform user experience▫ More cell-edges created▫ Near-far effect▫ Closed-access femtocells in co-channel deployments▫ Relay stations may have different duplexing schedulesH d ff d i i h t id b kh l it• Handoff decisions have to consider backhaul capacity, especially for relays and femtocells

• Optimized use of radio resourcesOptimized use of radio resources• Inter-cell interference management▫ Lack of coordination between cells▫ Scalability, security and limited backhaul bandwidth


FemtocellsFemtocells• Femtocells are low-power wireless access points that operate in

li d t t t t d d bil d i tlicensed spectrum to connect standard mobile devices to a mobile operator’s network using residential DSL or cable broadband connections [Source: Femto Forum].


CoverageCoverage• Inter-cell interference creates dead spots where UE QoS

t b t dcannot be guaranteed. ▫ Location w.r.t. MBS

Path loss shadowing fading▫ Path loss, shadowing, fading• Minimum distance of an FAP from the MBS s.t. a femto

outage probability (OP) constraintoutage probability (OP) constraint• Maximum density of simultaneously transmitting co-

channel femtocells meeting a macro/femto OP constraint• Maximum allowed transmit power of FAPs s.t. a macro

OP constraintMi i i d t it f FAP t f t OP• Minimum required transmit power of FAPs s.t. a femto OP constraint


Coverage Analysisg y


System ModelSystem Model• OFDMA downlink of collocated spectrum-sharing

macrocell and closed-access femtocells▫ A central MBS covers a disc area with radius rM

Femtocells of radi s are randoml distrib ted on R2 as a▫ Femtocells of radius rF are randomly distributed on R2 as a spatial Poisson point process (SPPP) with a density of λF.

▫ NF femtocells per cell site on averageF p g▫ UF indoor UEs per femtocell, each located on cell edge▫ MBS transmit power is PM,Tx per RB▫ FAP transmit power is PF,Tx per RB▫ Each FAP transmits with a probability ρ within each RB.

S ti l i t it f i lt l t itti h l▫ Spatial intensity of simultaneously transmitting co-channel FAPs is uF = λFρ.


Channel ModelChannel Model• Each subchannel sees Rayleigh flat fading and lognormal

shadowingg• Path loss follows the IMT-2000 channel model ▫ MBS to outdoor UE

MMM

3-7.1MMM 10PL ααφ DfD ==

▫ MBS to indoor UE McMMM 10PL φ DfD

FMFMFMFM

3c

-7.1FMMFMFMFM 10PL ααα ξξφφ DfDD ===

▫ Home FAP to indoor UE

▫ FAP to outdoor UEFF

F7.3

FFF 10PL ααφ DD ==

FAP to outdoor UE

▫ Interfering FAP to indoor UE

MFMFMFFMFMFMFPL αα ξφφ DD ==

2

▫ ξ denotes wall-penetration lossFFFF

FF2

FFFFFFFPL αα ξφφ DD ==


Femtocell DL SIRFemtocell DL SIR• For a given RB, the received SIR at an FUE is

1

∑Φ∈

−−−−

−−

+=

iiii DQHPDQHP

rQHPFFFM

F

FFFFFF1

FFFFMFMFM1

FMM

FFF1

FFFSIR αα

α

φφφ

▫ PF = PF,TxGFAPGUE, PM = PM,TxGMBSGUE;▫ DFM is the distance from the MBS to the FUE, DFFi is the distance

f i t f i FAP t th FUE

i

from interfering FAP i to the FUE; ▫ αF, αFM and αFF are path loss exponents from the home FAP, the

MBS and an interfering FAP to the FUE, respectively;g p y▫ HF, HFM and HFFi are unit-mean exponential channel power gains;▫ QF ~ LN(ζμF, ζ2σF

2), QFM ~ LN(ζμFM, ζ2σFM2) and QFFi ~ LN(ζμFF, ζ2σFF

2)denote lognormal shadowing ζ 0 1l 10denote lognormal shadowing, ζ = 0.1ln10;

▫ Φ is the set of FAPs transmitting in the given RB, with intensity uF.


Macrocell DL SIRMacrocell DL SIR• Co-channel interference from neighboring macrocells is ignored.

S• For a given RB, the received SIR at an MUE is

∑ −−

−−

=DQHP

DQHPMF

M

1MMM

1MM

MSIR α

α

φφ

▫ DM is the distance from the MBS to the MUE, DMFi is the distance from FAP i to the MUE;

∑Φ∈i

iii DQHP MFMFMFMFMFFφ

i to the MUE;▫ αM and αMF are path loss exponents from the MBS and an FAP to the MUE;▫ HM and HMFi denote unit-mean exponential channel power gains;▫ QM ~ LN(ζμM, ζ2σM

2) and QMFi ~ LN(ζμMF, ζ2σMF2) denote lognormal

shadowing.


Femto Outage ProbabilityFemto Outage Probability• Outage probability of an FUE w.r.t. the target SIR γF

⎞⎛( )

⎟⎟⎟

⎠

⎞

⎜⎜⎜

⎝

⎛<

+=<

∑Φ

−− FFFFFFF

1FFFFM

FFF PSIRP

FFγ

φγ αDQHPI

S

iiii

⎟⎟⎠

⎞⎜⎜⎝

⎛≥<+⎟⎟

⎠

⎞⎜⎜⎝

⎛<=

⎠⎝ Φ∈

FFM

FFFF

FM

F ,SIRPP γγγIS

IS

i

• Based on the stochastic geometry theory and for an FUE at a distance dFM from the MBS,

⎠⎝⎠⎝ FMFM II

⎞⎛( )

⎫⎧

+⎟⎟⎠

⎞⎜⎜⎝

⎛+−≈=<

dPrPFdD 2

FM2FFMF

FMFMF

FFFMFMFMFF

~~,~~;SIRP FM

F

α

α

ϑ σσμμφ

γφγ

( )( )( ) ( )∑∑

−+++

⎪⎭

⎪⎬⎫

⎪⎩

⎪⎨⎧

⎥⎥⎦

⎤

⎢⎢⎣

⎡−−−

N M

b

bbamn

mmn eeuvw

~

2~2

F~~22

FFexp1

FFFMFMFF αμσμσχ γκ((


( ) ( )∑∑= = +n m

bmn

meba1 1~2 χχπ

Macro Outage ProbabilityMacro Outage Probability• Outage probability of an MUE w.r.t. the target SIR γM

⎞⎛( )

⎟⎟⎟

⎠

⎞

⎜⎜⎜

⎝

⎛<=<

∑ −− MMFMFMF

1MFF

MMM MF

PSIRP γφ

γ αiii DQHP

S

• Based on the stochastic geometry theory and for an MUE at a distance dM from the MBS

⎟⎠

⎜⎝∑Φ∈i

distance dM from the MBS,

( ) ∑⎥⎥⎦

⎤

⎢⎢⎣

⎡⎟⎟⎠

⎞⎜⎜⎝

⎛−−−−≈=<

Mmm buvdD

1 MF

M

MF

MFMMMMM

2~22expexp1SIRPαμ

ασκ

πγ

(

= ⎥⎦⎢⎣ ⎠⎝m 1 MFMF ααπ

⎟⎟⎞

⎜⎜⎛

+⎟⎟⎞

⎜⎜⎛ 2

MFMF

2

MF~2~2exp

MF σμγπκαP

⎟⎟⎠

⎜⎜⎝

+⎟⎟⎠

⎜⎜⎝

= 2MF

MF

MF

MF

MF

MFM exp

ααφπκ


Minimum MBS-to-FAP DistanceMinimum MBS to FAP Distance• For an FUE at a distance dFM from the MBS,

⎟⎟⎠

⎞⎜⎜⎝

⎛<=⎟⎟

⎠

⎞⎜⎜⎝

⎛=<

FMFMF

FFFM

FMFM

FFFMFMF

FM

F PPFM

F

φγφγ α

α

dPrP

QHQHdD

IS

ϑ H Q /(H Q ) lognormal distribution

⎟⎟⎠

⎞⎜⎜⎝

⎛+−≈ 2

FM2FFMF

FMFMF

FFFM ~~,~~; FM

F

σσμμφ

γφα

α

ϑ dPrPF

▫ ϑ = HFQF/(HFMQFM) ~ lognormal distribution• Minimum dFM for P(SF/IFM < γF|DFM = dFM) ≤ εF,

FM

1221 ~~~~ αφ

−⎤⎡ ⎟⎞⎜⎛ FM

FFFFM

2FM

2FFMFF

1FMF

minFM,

~~,~~; α

α

ϑ

γφ

σσμμεφ −

⎥⎥⎥

⎦

⎤

⎢⎢⎢

⎣

⎡ ⎟⎠⎞⎜

⎝⎛ +−

≈rP

FPd

• Any UE less than dFM,min from the MBS should be associated with the macrocell.

⎦⎣


Maximum Femto DensityMaximum Femto Density• Maximum intensity of simultaneous co-channel femtocell

t i i t di t d ( ) f th MBS f P(SIRtransmissions at a distance dM (≤ rM) from the MBS for P(SIRM < γM|DM = dM) ≤ εM

( ) ( )MFM1

SIRMF ,, dPFdu ε−Δ=( ( ) ( )MFMSIRMF ,,

M

( ) ( ) MMMMMMFFSIR SIRP,,M

εγ ==<= dDdPuF

• Maximum effective femtocell density at a distance dFM (≥ dFM,min) from the MBS for P(SIRF < γF|DFM = dFM) ≤ εF

( ) ( )1~ dPFd −Δ

( ) ( )FMFF1

SIRFMF ,,F

dPFdu ε=

( ) ( ) FFMFMFFFMFFSIR SIRP,,F

εγ ==<= dDdPuF

• For dFM,min ≤ d ≤ rM, ( ) ( ) ( ){ }dududu FFF~,min (≤


FAP Transmit PowerFAP Transmit Power• Maximum allowed PF at a distance dM (≤ rM) from the MBS s.t.

P(SIR |D d )P(SIRM < γM|DM = dM) ≤ εM

( ) ( )MFM1

SIRMF ,,M

duFdP ε−Δ=

(

• Minimum required PF at a distance dFM (≥ dFM,min) from the MBS s.t. P(SIRF < γF|DFM = dFM) ≤ εF

At a distance d (d ≤ d ≤ ) from the MBS

( ) ( )FMFF1

SIRFMF ,,~F

duFdP ε−Δ=

• At a distance d (dFM,min ≤ d ≤ rM) from the MBS, ▫ if , then ; ▫ otherwise, no femtocell coverage and have to reduce uF = λFρ.

( ) ( )dPdP FF~ (

≤ ( ) ( ) ( )dPdPdP FFF~ (

≤≤otherwise, no femtocell coverage and have to reduce uF λFρ.


Simulations and Results


Simulation SetupSimulation Setup• One MBS in the center

FAP d MUE d l d d ithi th ll• FAPs and MUEs are randomly dropped within the macrocell coverage following independent SPPPs.

• ξ = 5 dB 10 dB • PM T = 43 dBmξ 5 dB, 10 dB• αM = 4• αF = 3• α = α

PM,Tx 43 dBm• PF,Tx ≤ 23 dBm• GMBS = 15 dBi • G = 2 dBi• αFM = αM

• αFF = 3.5• αMF = αFF

8 dB

• GFAP = 2 dBi• GUE = 0 dBi• rM = 1000 m

30• σM = 8 dB • σF = 4 dB • σFF = 12 dB

10 dB

• rF = 30 m• UF = 2• γM = 3 dB

5 dB• σMF = 10 dB• σFM = 10 dB• fc = 2000 MHz

• γF = 5 dB• εM = εF = 0.1• M = N = 12


Outage ProbabilityOutage Probability1

PM,Tx

= 43dBm, PF,Tx

= 23dBm, ξ = 10dB

OP

M, 300m, simulation

0.8

0.9

OPM

, 300m, simulation

OPM

, 300m, formula

OPF, 300m, simulation

OPF, 300m, formula

OPM

, 600m, simulation

0.6

0.7

babi

lity

M

OPM

, 600m, formula

OPF, 600m, simulation

OPF, 600m, formula

0.4

0.5

Out

age

Pro

ba

0.2

0.3

0 10 20 30 40 50 60 70 80 90 1000

0.1

Number of Co−Channel Femto Transmissions per Cell Site


Number of Co−Channel Femto Transmissions per Cell Site

Minimum MBS-to-FAP DistanceMinimum MBS to FAP Distance500

PM,Tx

= 43dBm, OPF ≤ 0.1

ξ = 5 dB, simulation

450m

toce

ll (m

)

ξ = 5 dB, simulationξ = 5 dB, formulaξ = 10 dB, simulationξ = 10 dB, formula

350

400

acro

BS

and

Fem

to

300

ance

bet

wee

n M

ac

200

250

Min

imum

Dis

tanc

13 14 15 16 17 18 19 20 21 22 23150

200

FAP Transmit Power (dBm)

M


FAP Transmit Power (dBm)

Maximum Femto Density Maximum Femto Density 10

3

PM,Tx

= 43dBm, PF,Tx

= 23dBm, OPM

≤ 0.1

ξ = 5dB, simulation

er C

ell S

ite

ξ = 5dB, simulationξ = 5dB, formulaξ = 10dB, simulationξ = 10dB, formula

102

Tra

nsm

issi

ons

per

101

Cha

nnel

Fem

to T

r

101

Max

imum

Co−

Ch

200 300 400 500 600 700 800 900 100010

0

Distance from Macro BS (m)

M



Maximum Allowed PF TMaximum Allowed PF,Tx50

PM,Tx

= 43dBm, NF = 100, OP

M ≤ 0.1

ξ = 5dB, ρ = 1, simulation

40

)

ξ = 5dB, ρ = 1, simulationξ = 5dB, ρ = 1, formulaξ = 5dB, ρ = 0.1, simulationξ = 5dB, ρ = 0.1, formulaξ = 10dB,ρ = 1, simulationξ = 10dB, ρ = 1, formula

30

mit

Pow

er (

dBm

) ξ = 10dB,ρ = 0.1, simulationξ = 10dB, ρ = 0.1, formula

23 dBm

10

20

imum

FA

P T

rans

m

0

10

Max

im

200 300 400 500 600 700 800 900 1000−10




Minimum Required PF TMinimum Required PF,Tx35

PM,Tx

= 43dBm, NF = 100, OP

F ≤ 0.1

ξ = 5dB, ρ = 1, simulation

25

30)

ξ = 5dB, ρ = 1, simulationξ = 5dB, ρ = 1, formulaξ = 5dB, ρ = 0.1, simulationξ = 5dB, ρ = 0.1, formulaξ = 10dB, ρ=1, simulationξ = 10dB, ρ=1, formula23 dBm

15

20

mit

Pow

er (

dBm

) ξ = 10dB, ρ=0.1, simulationξ = 10dB, ρ=0.1, formula

23 dBm

5

10

15

mum

FA

P T

rans

m

0

5

Min

imu

200 300 400 500 600 700 800 900 1000−10

−5

Distance between Macro BS and Femtocell (m)


Distance between Macro BS and Femtocell (m)

PF T under Low AttenuationPF,Tx under Low Attenuation40

PM,Tx

= 43dBm, NF = 100, ξ = 5dB

max P

F st OP

M ≤ 0.1, ρ = 1

30

35 (

dBm

)

max PF st OP

M 0.1, ρ = 1

min PF st OP

F ≤ 0.1, ρ = 1

max PF st OP

M ≤ 0.1, ρ = 0.1

min PF st OP

F ≤ 0.1, ρ = 0.1

max PF st OP

M ≤ 0.1, ρ = 0.5

20

25

Tra

nsm

it P

ower

(d max P

F st OP

M 0.1, ρ = 0.5

10

15

/Min

imum

FA

P T

ra

0

5

Max

imum

/M

200 300 400 500 600 700 800 900 1000−10

−5




PF T under High AttenuationPF,Tx under High Attenuation50

PM,Tx

= 43dBm, NF = 100, ξ = 10dB

max P st OP ≤ 0.1, ρ = 1

40dB

m)

max PF st OP

M ≤ 0.1, ρ = 1

min PF st OP

F ≤ 0.1, ρ = 1

max PF st OP

M ≤ 0.1, ρ = 0.1

min PF st OP

F ≤ 0.1, ρ = 0.1

max PF st OP

M ≤ 0.1, ρ = 0.5

30

rans

mit

Pow

er (

dB max PF st OP

M ≤ 0.1, ρ = 0.5

10

20

/Min

imum

FA

P T

ra

0

10

Max

imum

/M

200 300 400 500 600 700 800 900 1000−10




ConclusionsConclusions• OFDMA downlink of collocated spectrum-sharing

ll d l d f t llmacrocell and closed-access femtocells▫ Analytical expressions for outage probabilities

• It is possible to improve coverage by• It is possible to improve coverage by ▫ regulating femtocell transmit powers, which depend on the

distance from the MBS;▫ restricting the probability of each femtocell transmitting in

each RB, which can be controlled in both frequency and time domains.


Future WorkFuture Work• Mechanism for an FAP to infer its distance from the

l t MBS th t th FAP d t it t itclosest MBS, so that the FAP can adapt its transmit power accordingly to ensure satisfactory coverage.

• Associate UEs to cells intelligentlyAssociate UEs to cells intelligently▫ UEs associated to a cell with least path loss▫ Allow more UEs to benefit from low-power BSs

• Distributed adaptive resource partitioning▫ BSs negotiate resource reservation with each other▫ Resource request/grant messages sent over backhaul ▫ Based on load status and feedback from active UEs

C di t d lti i t t i i i th h t• Coordinated multi-point transmission in the heterogeneous network


Thank You !Thank You !Thank You !Thank You !Xiaoli Chu

E-mail: [email protected]


Coverage in Heterogeneous Networks · Femtocells • Femtocells are low-power wireless access...

Documents

Transcript of Coverage in Heterogeneous Networks · Femtocells • Femtocells are low-power wireless access...