OI´s Case Study Alberto Boaventura

2014-04-30

Changes and Challenges

TELECOMM IS BECOMING MOBILE MOBILE IS BECOMING DATA DATA IS BECOMING VIDEO VIDEO IS BECOMING SOCIAL

0

200

400

600

800

1.000

2009 2010 2011 2012 2013

SmartphonesTabletsNetbooksNotebooksDesktops

Source: Morgan Stanley & Nomura 2012

Wo

rld

De

vice

Sh

ipm

en

ts (

Mill

ion

s)

Source: Ericsson 2013 2009 2010 2011 2012 2013

1000

1800

Voice

Data

Tota

l (U

L+D

L) t

raff

ic (

Pe

taB

yte

s)

Source: Cisco VNI 2012

12

2012 2013 2014 2015 2016 2017

6

Mobile File Sharing

Mobile M2M

Mobile Web/Data

Mobile Video

Exab

yte

s p

er

mo

nth

In 2016, Social Newtorking will be second highest penetrated consumer mobile service

with 2, 4 billion users – 53% of consumer mobile users - Cisco 2012

2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

10

6

LTE UMTS/HSPA GSM;EDGE TD-SCDMA CDMA Other

Wo

rld

Mo

bile

Su

b. (

Bill

ion

s)

Source: Ericsson 2012

Voice Centric

Data Centric

Traffic

Reveue

1

2 34

5

RAPID LIFE CYCLE M2M, NEW DEVICES & APPS. CUSTOMER EXPERIENCE TRAFFIC & REVENUE DECOUPLING

Changes and Challenges

ITU-R M.2078 projection for the global spectrum requirements in order to accomplish the IMT-2000

future development, IMT-Advanced, in 2020.

531 MHz 749 MHz

971 MHz

749 MHz

557 MHz 723 MHz

997 MHz

723 MHz

587 MHz 693 MHz

1027 MHz

693 MHz

Region 1 Region 2 Region 3

MORE SPECTRUM NEW TECHNOLOGY SPLIT CELL

𝑪 𝒃𝒑𝒔 ≤ 𝑩(𝑯𝒛) ∙ 𝒍𝒐𝒈𝟐 𝟏 + 𝑺𝑰𝑵𝑹

Smallcells

Heterogeneous Network

hnm

h21

h12

h11

Mobile operation needs spectrum below 6 GHz,

but there is no enough around world.

Interference with exiting services: cleanup cost,

interference mitigation

High spectrum cost: The average license cost in

new spectrum auctions ranges around 100-700

million of Reais per 10 MHz FDD block

Spectrum Refarming

Spectral Efficiency

New infrastructure investment

Technology life cycle and adoption

Market Scale

New site legal barriers

Tax barriers

New site investment

Interference control and mitigation

Backhaul capillarity

MAIN BARRIERS

Carrier Aggregation

High Order MIMO

Cell Site Densification

Equation:

Data

Voice 2G (GSM, GPRS, EDGE)

3G (UMTS, HSPA+)

900 MHz (B8)

1800 MHz (B3)

2100 MHz (B1)

850 MHz (B5)

The Mobile Operation Planning involves the assessment of the complex equation:

Service (demand characteristics for voice and data) vs Technology (2G, 3G and 4G or otherwise) vs Spectrum

(900, 1800, 2100, 2600 MHz or otherwise), where should seek cost optimization not only present but future

disruptive scenario with lack of fundamental resource: spectrum.

Service Technology Spectrum

2600 MHz (B7)

700 MHz (B28)

450 MHz (B31)

4G (LTE)

CDMA/TDMA

Technology Life Cycle Ecosystem

Total Cost Ownership Customer Experience

Terminal Penetration & Cost Capacity&Spectral Efficiency

Service Support Level of Terminal Subsidy

License & Network Cost License Obligation

Ecosystem Bandwidth Limitation Coverage & Capacity

Interference Level of Terminal Subsidy

Constrains &

Decision Criteria

How to Solve?

Planning Framework

Data Voice

2G 3G 4G (LTE)

900 MHz 1800 MHz 2100 MHz 2600 MHz

Other

Data Voice

2G 3G 4G (LTE)

900 MHz 1800 MHz 2100 MHz 2600 MHz

VISION DEFINITION FRAMEWORK DEFINITION

• Service characteristics requirements, traffic requirements

Demand Analysis

• Network service assessment, Capacity evaluation, System growth opportunity, Split cell vs interference, Spectrum availability, License obligation, New technologies

System Analysis

• Technology life cycle, Ecosystem analysis, Spectral efficiency

Technology Analysis

• Scenario options, Total Cost Analysis, Spectrum availability, License obligation,

Scenario Analysis

Network Planning

Demand

Voice & Data

𝒎𝒊𝒏 𝑻𝑪𝟏 𝒊

(𝟏 + 𝑲)𝒊−𝑿, ⋯ ,

𝑻𝑪𝑵 𝒊

(𝟏 + 𝑲)𝒊−𝑿

𝑵

𝒊=𝑿

𝑵

𝒊=𝑿

⇒ 𝐓𝐚𝐫𝐠𝐞𝐭 𝐒𝐜𝐞𝐧𝐚𝐫𝐢𝐨

Scenario 1

Scenario N

...

New Frequency New Technology New Site

Long term scenario

Required for service and technology evolution

Required for long term spectrum management

Service and Strategic Needs

Plan Acquire Maintain Renew/ Dispose

Capacity, Technology & Spectrum Management

Traffic Demand Density

#𝑺𝒕𝒂𝒕𝒊𝒐𝒏𝒔 = 𝑴𝒂𝒙 𝑪𝒐𝒗𝒆𝒓𝒂𝒈𝒆; 𝑪𝒂𝒑𝒂𝒄𝒊𝒕𝒚

𝑪𝒐𝒗𝒆𝒓𝒂𝒈𝒆 =𝑨

𝑨𝒄

Where: A: Coverage Area Ac: Base Station Coverage Area D: Traffic Demand Ct: Base Station Traffic Capacity

𝑪𝒂𝒑𝒂𝒄𝒊𝒕𝒚 =𝑫

𝑪𝒕

𝑨

𝑨𝒄

𝑪𝒐𝒗𝒆𝒓𝒂𝒈𝒆><

𝑪𝒂𝒑𝒂𝒄𝒊𝒕𝒚

𝑫

𝑪𝒕

𝑪𝒕

𝑨𝒄

𝑪𝒐𝒗𝒆𝒓𝒂𝒈𝒆><

𝑪𝒂𝒑𝒂𝒄𝒊𝒕𝒚

𝑫

𝑨 D/A: Traffic Demand Density (Traf/km2)

Ct/Ac: System Offering Dens. (Traf/km2)

System Offering Density

𝑫

𝑨>𝑪𝒕

𝑨𝒄

Capacity Investment

𝑫

𝑨<𝑪𝒕

𝑨𝒄

Higher Cell Range, lower investment

level

𝑫

𝑨=𝑪𝒕

𝑨𝒄

Optimized Investment

Yes

No

Yes

No

Yes

Traffic

Coverage

Cap

acit

y More Spectrum

New technologies

Split Cells

Cell Range

Exceeded Traffic 𝑨𝒄

𝑪𝒕

System Rural Suburban Urbao

GSM 1800 MHz (5) 0,1 Erl/km2 3,2 Erl/km2 45,3 Erl/km2

UMTS 2100 MHz (5) 8,0 Erl/km2 41,7 Erl/km2 264,2 Erl/km2

HSPA+ 2100 (10) 10,4 Mbps/km2 21,5 Mbps/km2 35,3 Mbps/km2

LTE 700 MHz (10) 1,4 Mbps/km2 3,0 Mbps/km2 5,4 Mbps/km2

LTE 1800 MHz (10) 9,2 Mbps/km2 19,4 Mbps/km2 32,3 Mbps/km2

LTE 2600 MHz (10) 16,4 Mbps/km2 33,6 Mbps/km2 53,3 Mbps/km2

LTE 2600 MHz (20) 32,8 Mbps/km2 67,2 Mbps/km2 106,6 Mbps/km2

SmallCell 2600 MHz (10) 8584,7 Mbps/km2

SmallCell 2600 MHz (20) 17169,3 Mbps/km2

𝑪𝒕

𝑨𝒄

𝑫

𝑨

Coverage

Capacity Investment

↓

↑

0,0 Mbps/km2

300,0 Mbps/km2

600,0 Mbps/km2

900,0 Mbps/km2

0,3 km0,4 km0,5 km0,6 km0,7 km

2014

2016

2017

2018

2019

2020

Demands

2015

Capacities Cat 3

(initial) Cat 4 Cat 3

+ 5 MHz Cat 3

+ 10 MHz CA

Coverage

A

C B D E

250%

100%

73%

24%

F

Capacity, Technology & Spectrum Management

DATA: 4G CASE VOICE: 2G/3G CASE

Associated TCO TC1 TC2 TC3 TC4 TC5

700 Erl/km2

800 Erl/km2

900 Erl/km2

1000 Erl/km2

0,3 km0,4 km0,5 km0,6 km0,7 km

Coverage

2014

2016

2017

2018 2019

2020

Demands

2015

Capacities 2G (100 T)

3G (1 C) 2G (50 T) 3G (2 C)

2G (25 T/V) 3G (2 C) 3G (3 C)

A

C

D E B

- 5 MHz

- 5 MHz

27%

Associated TCO TC1 TC2 TC3 TC4

0,0 Mbps/km2

300,0 Mbps/km2

600,0 Mbps/km2

900,0 Mbps/km2

2014 2015 2016 2017 2018 2019 2020

Demand Capacity

A

G

H

F

Cat 3 + 5 MHz

Cat 3 + 10 MHz

CA

Cat 3 (initial)

0 Erl/km2

300 Erl/km2

600 Erl/km2

900 Erl/km2

1200 Erl/km2

2014 2015 2016 2017 2018 2019 2020

Demand Capacity

A C

D

2G (100 T) 3G (1 C)

2G (50 T) 3G (2 C) - 5 MHz

2G (25 T/V) 3G (3 C) - 10 MHz

G

H

Spectral Efficiency

0 1 2 3 4 5 6 7

200kHz

25 TRX

3,84MHz

1 WCDMA Carrier

r

R

D

i j

i

j D

r

R

D

i

j

i

jD

Codec FR D = 4 / Sector = 3

Reuse = 4 x 3 #Ckt/Sector= 2x7=14

Codec AMR 12.2 127 Walsh Codes

Reuse = 1 %SHO=20%

#Ckt/Sector = 64

24 Erl/BTS 160 Erl/NodeB

r

R

D

i

j

i

jD

PRBs

...

7 S

ymb

ols

12 subcarriers

25 Resource Blocks

700 Erl/eNB Codec AMR 12.2

25 PRBs - 300 REs 200 -250 users/ Sector

2G (GSM) 3G (UMTS/HSPA) LTE

HSPA+ 2100 MHZ VS LTE 2600 MHZ 3G (UMTS/HSPA) LTE

Voice Capacity @ 5 MHz

Data Capacity @ 5 MHz

Source: Brendan McWilliams, Yannick Le Pézennec, Grahame Collins Vodafone Technology Networks, Access Competence

Centre, Madrid, Spain & Newbury, United Kingdom 2012

𝑻𝒉𝒓 = #𝑪𝒐𝒅𝒔 ×𝑴𝒐𝒅 × 𝑭𝑬𝑪 ×𝐶ℎ𝑖𝑝𝑅𝑎𝑡𝑒

𝑺𝑭

𝑻𝒉𝒓 = 𝟏𝟓 × 𝟔 × 𝟏 ×𝟑, 𝟖𝟒

𝟏𝟔= 𝟐𝟏 𝑴𝒃𝒑𝒔

𝑻𝒉𝒓 = #𝑴𝑰𝑴𝟎 × #𝑹𝑩𝒔 ×𝑴𝒐𝒅 × 𝑭𝑬𝑪 ×#𝑪𝒂𝒓.× #𝑺𝒚𝒎𝒃

𝑻𝑻𝑰/𝟐

𝑻𝒉𝒓 = 𝟐 × 𝟐𝟓 × 𝟔 × 𝟏 ×𝟏𝟐 × 𝟔 − 𝟏𝟐

𝟎, 𝟓= 𝟑𝟔 𝑴𝒃𝒑𝒔

MIMO Yes, but not for existing network

Modulation QPSK, 16 QAM, 64 QAM

Intereference Rake Receiver

Limitation Up Link limitation due interference

MIMO Yes,

Modulation QPSK, 16 QAM, 64 QAM

Intereference FRF/ICIC

Limitation CoMP/ICIC/e-ICIC

Hundreds of users per NodeB Thousands of users per eNB

Cost

$$$

$$$

$$$

$$$

3G (1) 3G (2) 3G (3) 3G (4)

3a. Portadora

2a. Portadora

NodeB

LTE (X)

R99 HSDPA + Carriers HSPA+ DC MIMO?

FLAT ARCHITECTURE LTE VS 3G BASE STATION COST NETWORK UPGRADE

S1-U

X2 S11

PCRF

HLR/HSS

OCS/ OFCS

Internet S5

S-GW P-GW

MME

IMS

S1-AP

Gx

Rx S6a

SGi

Gy/Gz Sy

Ro/Rf Sh

Sp

E-UTRAN Evolved Packet Core

HSPA needs continuous upgrade for supporting high data throughput/capacity

LTE/SAE has flat architecture without controllers (like RNC) and CS domain.

LTE Access Network is 7-10 times cheaper than 3G per Mbps.

NEW CELL SITE CAPEX SPECTRUM COST NETWORK COST

25% 45% 50%

52% 38% 35%

23% 17% 15%

Rooftop 30m Tower 50m Tower

Infra BTS Transport

Source: Planning Area, Oi, 2012

New Cell Site represents a huge impact in Wireless Operation total cost.

System capacity (Spectral Efficiency) in single site is the most important attribute.

The 2G spectrum consumption is faster than 3G as voice traffic increases. Spectrum is a lack and valuable resource. 10 MHz can cost

500-1 Billion of Reais.

0 MHz

10 MHz

20 MHz

30 MHz

10 Erl/BTS 50 Erl/BTS 90 Erl/BTS 130 Erl/BTS

2G

3G

+14 MHz

Voice Perspective

Data Perspective

$$$

$$$

$$$

$$$

$$$

0,0 kErl 2,0 kErl 4,0 kErl 6,0 kErl

2G (4/4/4)

3G (1/1/1)

3X

The cost per Erl to support voice on 3G is invariably cheaper than 2G .

6 kErl (~ 300k users), is 3 x the cost 2G.

Although LTE and HSPA have almost the same spectral efficiency, but…

+$ +$ +$ +$

21/28 42 84 168 336

150/300 1500

Rel.7 Rel.8 Rel.9 Rel.10 Rel.11 HSPA+

LTE

Mbps

Mbps

Customer Experience

THROUGHPUT

0 Mbps

2 Mbps

4 Mbps

6 Mbps

2009201020112012201320142015

América Latina

America do Norte

Europa Ocidental

Brazil

It is expected that the average grows exponentially. In Brazil, the growth is

82% year-on-year by 2015 according to Cisco

APPS AND MARKET TRENDS

URA

DCH

PCH

FACH Idle

3G: RRC SM (3GPP TS 25.331)

LTE: RRC SM (3GPP TS 36.331)

Idle

Conn.

LATENCY VOICE QUALITY

QoE is the main motivation of churn and it will remain a key challenge for mobile operators and may in fact rise as the

wireless value chain becomes increasingly decentralized.

ITU-T Rec. P.10/G.100: The overall acceptability of an application or service, as perceived subjectively by the end-user.

QOE DEFINITION

Req. SLA QoS

QoE SLA KQI KPI

0 ms 400 ms 800 ms

GPRS

EDGE

WCDMA

HSPA

HSPA+

LTE

2

3

4

4 kbps 11 kbps 18 kbps

AMR-NB AMR-WB

21/28 42 84 168 336

150/300 1500

Rel.7 Rel.8 Rel.9 Rel.10 Rel.11 HSPA+

LTE

Mbps

Mbps ~10 ms

t t+ t- throughput

u(t)

u(t+)

u(t-)

u”(t) <0

Utility=QoE

Utility function perfectly captures user satisfaction in terms of what they are

willing to expect and pay.

UTILITY FUNCTION VS QOE

Users have more sensibility when lose than when win.

HDTV Video Conf.

Aug. Reality Collaboration

Mobile TV Music

Email

M2M

Games

Social Net.

SMS IM

kbps Mbps Mbps+

Other improvements, such as HD Voice (or High-Definition Voice) represents a significant customer satisfaction through high sound quality in mobile

communications.

Competitive Pressure

Delighted

Extremely Dissatisfied

Fully Functional Dysfunctional

Attractive

Time Expected

KANO´S MODEL

Customer satisfaction has a positive correlation with how the product is functional. I.e., dysfunctional => bad experience; functional => Delighted

Technology Life Cycle

2012 2013 2019E CAGR

2013-2019

Worldwide mobile subscriptions* 6,300 6,700 9,300 6%

– Smartphone subscriptions 1,300 1,900 5,600 20%

– Mobile PC, tablet &mobile router subs 250 300 750 15%

– Mobile broadband subscriptions 1,500 2,100 800 25%

– Mobile subscriptions, GSM/EDGE-only 4,300 4,300 1,200 -20%

– Mobile subscriptions, WCDMA/HSPA 1,200 1,600 4,800 20%

– Mobile subscriptions, LTE 70 175 2,600 55%

Source: Ercisson Mobility Report 2013

2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

10

6

LTE UMTS/HSPA GSM;EDGE TD-SCDMA

CDMA Other

Wo

rld

Mo

bile

Su

bs.

(B

illio

ns)

Source: Ericsson 2012

LIFE CYCLE SUBSCRIPTIONS RATE AND FORECAST

Utility Budget Restriction

LTE

HSP

A+

𝑴𝑹𝑺 =

𝝏𝑼𝝏𝑳𝑻𝑬𝝏𝑼

𝝏𝑯𝑺𝑷𝑨 +

⇒ 𝑴𝑹𝑺 =𝟏. 𝟓𝟓

𝟏. 𝟑𝟑= 𝟏. 𝟐𝟗

𝒑𝑳𝑻𝑬𝒑𝑯𝑺𝑷𝑨+

≤ 𝑴𝑹𝑺 ⇒ 𝒑𝑳𝑻𝑬 ≤ 𝟏. 𝟐𝟗 × 𝒑𝑯𝑺𝑷𝑨+

Utility Function

Budget Restriction

Per

form

ance

leve

l

time Will

ingn

ess

to p

ay

∆ 𝑾𝒊𝒍𝒍𝒊𝒏𝒈𝒏𝒆𝒔𝒔 𝑻𝒐 𝑷𝒂𝒚

∆ 𝑻𝒊𝒎𝒆=∆ 𝑻𝒉𝒓𝒐𝒖𝒈𝒉𝒑𝒖𝒕

∆ 𝑻𝒊𝒎𝒆 ×∆ 𝑾𝒊𝒍𝒍𝒊𝒏𝒈𝒏𝒆𝒔𝒔 𝑻𝒐 𝑷𝒂𝒚

∆ 𝑻𝒉𝒓𝒐𝒖𝒈𝒉𝒑𝒖𝒕

Throughput

Co

st

Uti

lity

UTILITY FUNCTION VS BUDGET RESTRICTION

Network Planning

Year X

Demand

Voice & Data

Scenario A Expand with existing technology

Scenario B Start to change to a new technology

𝑻𝑪𝑨 𝒊

(𝟏 + 𝑲)𝒊−𝑿

𝑵

𝒊=𝑿

𝑺𝒄𝒆𝒏𝒂𝒓𝒊𝒐 𝑩><

𝑺𝒄𝒆𝒏𝒂𝒓𝒊𝒐 𝑨

𝑻𝑪𝑩 𝒊

(𝟏 + 𝑲)𝒊−𝑿

𝑵

𝒊=𝑿

∆𝑻𝒆𝒓𝒎𝒊𝒏𝒂𝒍 𝑺𝒖𝒃𝒔𝒊𝒅𝒚 ≤ −∆𝑺𝒑𝒆𝒄𝒕𝒓𝒖𝒎 𝑪𝒐𝒔𝒕 − ∆𝑵𝒆𝒕𝒘𝒐𝒓𝒌 𝑪𝒐𝒔𝒕

Scen. A Scen. B

Year X

Ecosystem Cost

Spectrum Cost

𝑻𝑪𝑨 𝑻𝑪𝑩

ASSET MANAGEMENT

𝑬𝒄𝒐𝒔𝒚𝒔𝒕𝒆𝒎 𝑪𝒐𝒔𝒕 = 𝑵𝒆𝒕𝒘𝒐𝒓𝒌 𝑪𝒐𝒔𝒕 + 𝑻𝒆𝒓𝒎𝒊𝒏𝒂𝒍 𝑺𝒖𝒃𝒔𝒊𝒅𝒚

LTE VS HSPA LAUNCHED NETWORKS

20%

11%

7% 35%

1%

26% 0% <7,2 Mbps7,2 Mbps14,4 Mbps21,1 Mbps28,8 Mbps42,2 Mbps84,4 Mpbs

89%

9% 2% 100 Mbps

150 Mbps

300 Mbps

LTE: 279 networks

HSPA: 574 networks

Source: GSMA/GSA 2014

Final Words

Rapid and consistent mobile broadband consolidation, video, social network will bring a tsunami of IP

data traffic. In the same time rapid service lifecycle forces new services to be mature quickly, with low

profitability.

Churn is still remains a major challenge for operators, mainly due to costs associated with the capture

of new users in a consolidated market. Negative perception of services due smaller customer

experience failures and bad service performance are one of the major reasons for service cancellation

and change of service provider.

Definitely, the main application (or killer app) is customer experience.

Spectrum is a lack resource and ITU-R forecasts a need of 1280 to 1720 MHz more in the medium term

for IMT (before 2020) for Mobile Broadband Services.

New cell site represents most important part of network investment and has several legal barriers.

Technology evolution improves the overall customer experience, but also brings additional advantages

by optimizing and saving long term investments and spectrum.

Technology life cycle management is not simple, but it is one of the most important tool for strategic

network planning, but even for tactical and operational planning.

Alberto Boaventura alberto@oi.net.br +55 21 98875 4998

¡GRACIAS!

THANKS!

OBRIGADO!