Hereby my Broadband MEA 2013 keynote presentation (delivered 20 March 2013).It's an Extended Re-Mix, rather than the limited slides required for the 15 minutes talk. Hopefully this gives a more rounded view on Growth Dynamics related to mobile data demand and supply in mobile networks as well as the Economics involved (which is often forgotten when the usual suspects gets exited about x1,000 challenges).
Take also a peek at my Blog on the topic:
http://techneconomyblog.com/2012/11/02/the-economics-of-the-thousand-times-challenge/
Enjoy!
Btw. apologies for the fairly hopeless format conversion in this presentation ( it looks better in real ppt format ;-).
Transforming Data Streams with Kafka Connect: An Introduction to Single Messa...
Growth Pains: How mobile networks will supply data capacity for 2020
1. Growth Pains: How networks will
supply data capacity for 2020
Dr. Kim Kyllesbech Larsen
Head of Technology Office,
Group Technology.
Ooredoo
2. The New Maya’s?
2 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
3. Are you worried?
90+% GROWTH CELLULAR DATA GROWTH
MEA 176%! YoY
(2011-2012)
CELLULAR DATA GROWTH
MEA 130%! GAGR
(2011-2013)
Exponential
CELLULAR DATA GROWTH
MEA 93%! YoY
Growth!
(2012-2013)
CELLULAR DATA GROWTH
MEA 90%! CAGR
(2011-2017)
Should you be concerned?
Source: Cisco 2011 – 2017 Mobile
3 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
4. Super much more mobile data traffic
Illustration
MEA
2011 - 2016
Average Demanded Mbps per Sector
×2.5 versus Average Supplied Mbps
Date Revenues
3 TIMES MORE
SPECTRUM
EFFICIENCY SUPPLY
SITES
SUPPLY
×30
Data Volume
Note: assuming largely U900 based network grids with 70% load
and 2 10 – 2 15 MHz available. Average supplied throughput is cell
average. For MEA 2012 assumed to have 100 thousand Node-Bs
installed which is assumed doubled to 200 thousand by 2017.
Sources: Pyramid Research, Cisco VNI and various others.
4 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
5. Cisco VNI for MEA ... Penta1 byte society?
2012 = 50 PB 2017 = 860 PB
1 Million Blu-ray discs 17+ Million Blu-ray discs
per month per month
Per Node-B per Day:
at least
you should
be concerned!
Ca. 100 Thousand Node-Bs Min. 200 Thousand Node-Bs
Ca. ½ Blue-ray disc Ca. 4 Blue-ray discs
Average Node-B throughput 6 Mbps Average Node-B throughput 37 Mbps!!!!
1A Penta Byte is 1,000,000,000,000,000 (15 zeros) Bytes
Note: assuming double layer Blue-ray of 50GB capacity.
5 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
6. Growing pains...a sense of urgency?
MORE & BETTER SPECTRUM REQUIRED.
QUANTUM-LEAPING EFFECTIVE SPECTRAL EFFICIENCY.
MANY MORE SITES REQUIRED.
6 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
7. Growth ... the ugly, the bad and the good.
CAGR REFS TO YEAR 1 (THE NEW MAYA’s)
176%
YoY GROWTH RATE
DEMAND
SUPPLY
131%
ITS VERY NORMAL TO PRESENT
116%
A DEMAND GROWTH RATE 107%
99%
WITHOUT 91%
85%
79%
73%
- REFERENCE
- MENTION OF SUPPLY SIDE
THOUGH SUCH A NUMBER
IS IN ALL EFFECT
OF LITTLE MEANING. 176% 93% 90% 83% 68% 54% 53% 42% 35%
Yr1 Yr10
Be careful & critical when faced with growth rates ...
The New Maya’s might be at playing at your cognitive bias.
7 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
8. Growth ... is the sky the limit?
LIMITED?
?
Technology Usage
adaptation Adaptation
#Users Usage per User
8 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
9. Growth ... technology adaptation.
Population
Availability
Economics
LIMITED! 2017: 45+% of MEA1
Technology
adaptation
#Users 2013: ca. 30% of MEA1
on 3G or better
1 Pyramid Research MEA.
9 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
10. Growth ... usage adaptation.
Pricing
Use Cases
Cellular off-load
Convenience
CELLULAR TV
20 hrs. per week TV viewing Cloud
? Technology
Spectral capacity
Network Speed
@ 1Mbps unicast stream
Device performance
20 GB per Month per user Transport infrastructure
2017: 1.5 GB Cellular!
LIMITED?
Cellular Usage
Adaptation
Usage per User 2013: 300 MB Cellular!
per Month per User in MEA1
1 Cisco VNI MEA
10 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
11. MEA mobile data growth.
Illustration
MEA ~ to CEE 2017
Penta-Byte per month 40% of USA in 2017
2013 15% of APAC in 2017
<2011
2010
Cisco VNI
<2011
MEA
Pyramid Research 4Q12?
MEA 2Q12
2007
2Q11
2012
2Q10
2010
USA reach this
MB-level in 2011
Mobile broadband uptake in %. Mega-Byte per subs per month
Sources: Pyramid Research data until 2017 & Cisco VNI for MEA 2011 to 2017. Note: The Flags
illustrates what year a certain value was reached by that given country compared to MEA projections.
11 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
12. MEA growth & diffusion
UMTS will peak
at 50% penetration.
Billions of Subscriptions
50%
2G
2G remains for a
3G
(too) long time LTE
tying up
valuable spectrum!
Sources: Pyramid Research data until 2017; after 2017 authors own assessment based
on technology growth and diffusion dynamics.
12 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
13. Why MEA Growth might be different?
Illustration
PREPAID
GROWTH
DYNAMICS
13 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
14. Western Europe ... They have had it!
Illustration
2020
1.7 0.9
2012
2.1 1.9 0.91
1.6 0.8
2.33 5.4 81% <100%
PHH DHH FBB M2M
2008
2.0 1.2 0.47
1.5 0.7 2.36 4.1 63% ~10% Icon
PHH DHH FBB M2M
TV PC
1.9 0.4 0.00
Mobile subs Smartphone Tablet
2.37 2.6 51% <2% People Display HH with HH with
per HH per HH Fixed BB M2M
PHH DHH FBB M2M Subscr.
Note: HH stands for House Holds.
14 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
15. MEA ... We want connectivity!
Illustration 3+ number of Displays
8+ Smartphones increase 2020
0.33 0.9
Doubling of Displays
40 Smartphones increase
2012
4.8 3.3 0.5
0.30 0.6
4.66 5.0 5% <5?%
PHH DHH FBB M2M
2008
3.5 0.4 0.1
0.29 0.4 4.70 1.4 3% ~0% Icon
PHH DHH FBB M2M
TV PC
2.2 0.01 0.00
Mobile subs Smartphone Tablet
4.69 0.7 2% ca. 0% People Display HH with HH with
per HH per HH Fixed BB M2M
PHH DHH FBB M2M Subscr.
Note: HH stands for House Holds.
15 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
16. The mobile broadband journey!
Voice revenue decline (likely faster than data revenue uptake)
Messaging revenue decline (with Smartphones & OTT uptake)
Cash and margin pressure from new technology introduction.
Cash and margin pressure from sustaining data demand.
16 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
17. The Ugly: Telco-OTT Challenge ...
Illustration
LEGACY VOICE WEU
LEGACY MESSAGING
Between 2011 - 2021 WEU
MEA Between 2011 - 2021
Between 2011 - 2021
-30%
-50%
-15%
MEA
Between 2011 - 2021
+20%?
*
Sources: Pyramid Research (until 2017) and other analyst projections.
(*) It is hard to believe that SMS revenues will grow in MEA. Will depend a lot on Smartphone
penetration and mobile data deployment.
17 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
19. The Good: mobile data revenue growth ...
×180
WEU
Illustration Between 2011 - 2021
180 Million Subs
×2.5 On very dense networks
BUT MEA
Between 2011 - 2021
based on 1800/2100 grids
1.6 Billion Subs
On fairly coarse networks ×5
based on 900 grids
DATA VOLUME Comfortable & sustainable
Network Challenges (The Bad?) Network Basis for
coping with
to be overcome
Future Growth.
to cope with
Future Growth.
Sources: Pyramid Research (until 2017) and other analyst projections.
19 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
20. Capacity fundamentals
CAPACITY Ci = BANDWIDTH Bi × EFFICIENCY Ei × CELLS Ni
MHz Mbps per MHz per Cell #
Business as Usual New spectrum New technologies New macro
× (VERY) COSTLY BaU (COSTLY) VERY COSTLY
Innovation Re-farming Improvements Small-cells
× COMPLEX + EFFICIENT BaU (COSTLY) (VERY) COSTLY
Radical Spectrum sharing Spectrum sharing Site sharing
COMPLEX BUT EFFICIENT EFFICIENT
Leapfrog Network Capacity & Quality within Financial KPIs
20 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
21. Spectrum fundamentals
SPECTRUM ACQUISITION
SPECTRUM RETENTION
SPECTRUM PERFECTION
21 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
22. MEA state of mobile broadband 2020
CAPACITY Ci = BANDWIDTH Bi × EFFICIENCY Ei × CELLS Ni
MHz Mbps per MHz per Cell #
Demand in 2020 to 2012 ratio (Cisco VNI for MEA).
50 × (CAGR 63%)
Supply in 2020 to 2012 ratio
4–8× 1.5 × 2- 5×
Macro-based
9 – 60 × Low & high Incl. LTE, but
× ( Ksmall-cell >1) ×
frequency blend efficiency blend
Incl. small-cells
2012 Utilization Compensated Demand (i.e., directly comparable to Supply)
10 – 20 ×
2020 Mobile Broadband Networks in MEA likely to cope with expected demand
BUT ONLY WITH MUCH MORE SPECTRUM & MANY MORE SITES
NOTE: It should be well understood that there is a possible disconnect between demand
& supply and that to make the two comparable the utilization needs to be considered.
22 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
23. Spectrum acquisition & retention
.
23 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
24. Spectrum perfection
Illustration
LTE
TDD
UMTS
GSM
Mobile Data Mobile Data Mobile Data Capacity overlay &
10MHz 10MHz 75 (40) MHz Small Cells
NOT macro-option
4 – 7.5
+2600MHz (NEW)
+2100MHz (NEW)
1800MHz REFARM
900MHz REFARM
+800MHz (NEW)
TYPICAL OPERATOR SPECTRUM POSITION IN MEA
NOTE: I am assuming that E-GSM @ 900MHz has been implemented.
24 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
25. Spectrum re-farming and migration dynamics.
Illustration
Customer and Technology migration dynamics
100%
80%
LTE
60%
No 2G adds
40% 3G
20%
2G 2G buy-out
0%
TIME
Decline growth
Spectrum management dynamics
LTE 2G 3G 2G LTE 3G
800 MHz 900 MHz 1800 MHz 2100 MHz
NOTE: The above is not an emerging MEA example. Also not ethat spectral re-farming to UMTS
requires that there is multiples of 5MHz available.
25 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
26. Spectrum efficiency
MODERNIZATION
4
MULTIPLEXING GAIN
ANTENNA INNOVATION
CUSTOMER MIGRATION
YOUR NETWORK ON LATEST & GREATEST TECHNOLOGY ?
IF YOUR CUSTOMERS ARE NOT! … SO WHAT!
26 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
27. Need for (small) cells … Oh yes!
CAPACITY Ci = BANDWIDTH Bi × EFFICIENCY Ei × CELLS Ni
MHz Mbps per MHz per Cell #
2011 Simplified Illustration
Supplied (max)
2×10 MHz HSPA+ 42 DC Urban “hot” zone
DL ca. 10 Gbps @ 2.1GHz DL 0.97 / UL 0.34 1,000 Macro Cells1
UL ca. 3 Gbps
×15 ×4 LTE ×1.5 LTE ×2.5 LTE
2020
LTE Demand 2×20 MHz LTE 2x2 (cat. 3) Urban “hot” zone
@ 1800MHz DL 1.45 / UL 0.86 Up-to 2,500 Cells 2
DL 100 Gbps + 2x20 MHz +1,500 on-top
UL 30 Gbps @ 2.6GHz of exiting network
Small cells required!!!
CAGR ca. 35% 2 Femto-cell off-load!!!
1 Urban “hot” zone cell is defined as a cell with a range less than 500 meter,
2 Assuming 70% load as design rule.
27 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
28. MEA needs for MHz & Sites
How many new sites?
IF
Maximum 2 25 MHz 1
Economics cannot allow for
Eff: 0.8 – 1.5 Mbps/MHz/Cell these site numbers if macro!
Assuming Cisco VNI for MEA
Uncertainty 50k sites & 2 yrs. Demand for small cells and
alternative off-load techniques
How much additional MHz?
IF
Never more than 200K sites
Eff: 0.8 – 1.5 Mbps/MHz/Cell
Assuming Cisco VNI for MEA
Uncertainty 10 MHz & 2 yrs.
Realistic for operators to acquire
additional 2×40 MHz
Also highly likely to be blend of
low & high frequencies.
1 This could for example correspond to 2x10 MHz @ 900MHz & 2x15 MHz @ 2100MHz.
28 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
29. Economics of providing lots of capacity
BANDWIDTH Bi × EFFICIENCY Ei × CELLS Ni
MHz Mbps per MHz per Cell #
4–8× TCO 1.5 × TCO 2 – 5+ × TCO
• New Spectrum. +++ • Modernization. ++ • Macro Cell Splits. +++
• Spectral re-farming. n • Higher order MiMo. ++ • Small Cell ++
• Accelerated customer ++ • Beam forming (AAA). ++ deployment.
migration (re-farm). • Cognitive Radio. + • Cellular + WiFi +
• Carrier Aggregation. n • Carrier Aggregation. n hybrids.
• Spectrum Sharing. -- • Coordinated Multi- n • Off-loading to WiFi & -
Point (coMP) Tx/Rx. Femto-cells.
• Spectrum Sharing. -- • Network Sharing --
HIGH CAPEX (NEW SPTRM) BaU / HIGHER CAPEX HIGH TO VERY HIGH CAPEX
HIGH OPEX (MKT INVEST) LOW OPEX EXPOSURE. HIGH TO VERY HIGH OPEX
NOTE: TCO nomenclature is the following: (+++) VERY HIGH, (++) HIGHER, (+) HGH,
(n) neutral, (-) LOW TCO / POS NPV & (--) VERY LOW TCO / POS NPV.
29 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
30. Best place for small cells … at peoples home?
Illustration
Up-to 80% of all
cellular data traffic
generated in no
more than 3 cells.
data voice
00:00 6:00 8:00 10:00 12:00 14:00 17:00 22:00
Femto Cell Small Cells Femto Cell
@Home On the @ Work On the @Home
(1 – 2 Cells) Go (2 – 4 Cells) Go (2 – 3 Cells)
NOTE: WiFi is just a bridge to better cellular small-network systems become main stream
with controlable spectrum assets and E2E Customer Experience Management.
30 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
31. Economics of small cells? *
Extension of
existing infrastructure
more economical than
new Small-Cell Network
(if the choice is there!)
TCO Breakpoint 1
ca. 20 Small-cell Nodes
(compared to new macro-site!)
* Note: the above is based on WEU cost distribution. Differences are likely to occur for
other markets / regions, 1 Based on WEU cost structure.
31 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
32. Small cells scaling challenges - Capex
Illustration
Equipment
100 per unit <2 per unit 30k Macro Equipment
Core
transport 6k annualized Capex
equipment Feasible < 120 annualized Capex
+ antenna
? < 600 per unit Small-cell Node
civil works
Similarly
50 Node Small-cell CW should be
Macro-site
Small-Cell network < 1,200 per unit Small-cell Node 1
Annualized
Capex Annualized Capex
LARGE SCALE SMALL-CELL DEPLOYMENT REQUIRE
a QUANTUM-LEAP in ECONOMICS of scale & DEPLOYMENT INNOVATION.
1 this
is equivalent to no more than 60 hours of technical labor in US not considering
materials.
32 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
33. Small cells scaling challenges - Opex
Illustration
Rental:
100 per unit
energy <2 per unit Macro lease 10k require
Small-cell node unit lease to be
O&M no more than
transport
Feasible < 200 per anno!
rental
? Transport:
Macro transport 8.5k require
Small-cell node unit transport
Macro-site 50 Node
solution to be better than
Annual Small-Cell network
< 170 per anno.
Opex Annual Opex
OPEX SCALING is one of the BIGGEST CHALLENGES for
LARGE-SCALE SMALL-CELL NETWORK DEPLOYMENT.
33 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
34. What should we be passionate about
SPECTRUM PERFECTION, PERFECTION & PERFECTION.
EARLY SITE DENSITY PLANNING CRUCIAL FOR SUSTAINING GROWHT1
BE RADICAL ADAPTING INNOVATIVE CAPACITY ENHANCING STRATEGIES
1 See Refs by Iris Barcia & Simon Chapman (Keima Wireless, http://www.slideshare.net/zahidtg/radio-
challenges-and-opportunities-for-large-scale-small-cell-deployments
34 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
35. MEA 2020 cellular data challenges.
MOBILE BROADBAND NETWORKS WILL BE THE MAIN GATEWAY TO INTERNET.
LACK OF ECONOMICAL HIGH QUALITY TRANSPORT INFRASTRUCTURE
COULD SLOW DOWN BROADBAND GROWTH.
EVEN WITH SPECTRUM AVAILABILITY & PACED TECHNOLOGY MIGRATION
CELLULAR SITES WILL INCREASE DRAMATICALLY ACROSS MEA COMING YEARS.
35 March 19, 2013 Dr. Kim Kyllesbech Larsen, Broadband MEA 2013.
Maybe do a cognitive bias experiment related to growth rates?
Animation wise … initially I ll be showing 176% CAGR (nothing else). Then showing the demand curve, then the rest of the CAGR and the annual growth. Finally I ll show the Supply curve. Make the point that it has little point in just talking about a growth rate without specifying the reference point.