affordable & usable broadband: need to address...
TRANSCRIPT
Affordable & usable broadband: Need to address international
bottlenecks Rohan Samarajiva, Chair, LIRNEasia
Phil ICT Research Network, Manila, September 26, 2013
CAN ASIAN SUCCESS IN NARROWBAND BE REPLICATED WITH BROADBAND?
2
In voice+SMS basket comparison, 4 countries < USD 5 in 2007: All Asian
3
In 2009, 12 countries < USD 5: 9 were Asian
4
5
Success in voice: Low prices greater connectivity (India SEC D&E)
0
5
10
15
20
25
30
35
40
0
2
4
6
8
10
12
14
16
18
19
91
19
92
19
93
19
94
19
95
19
96
19
97
19
98
19
99
20
00
20
01
20
02
20
03
20
04
20
05
20
06
20
07
20
08
% o
f p
op
ula
tio
n
INR
per
min
ute
Mobile tariff (effective charge) Urban Indian SEC D and E mobile penetration
Rural Indian SEC D and E mobile penetration
This must be repeated with broadband
WHAT IS BROADBAND? IS BROADBAND DIFFERENT IN ASIA?
6
What is broadband?
• The Broadband Commission sought to focus on considering some of the core concepts of broadband as an always-on service (not needing to make a new connection to a server each time a user wants to go online), and high-capacity: able to carry lots of data per second, rather than the particular arrival speed of the data. The practical result is that broadband enables the combined provision of voice, data and video at the same time.
7
What is not broadband?
• Partnership for measuring the ICTs for Development suggests that always-on connections that are not 256kbps up and down are not broadband
• India, in its National Telecom Policy 2012, proposed gradually raising the threshold to 2Mbps down
• But what is the point of applying definitions to advertising, when real performance suggests that many operators in Asia do not offer true broadband (at least 256 kbps, 80% of the time)
8
BD offerings are not broadband? BSNL, Bangalore & Airtel, Chennai?
City, Country
Dhaka, BD Thimphu, BT
Bangalore, IN Chennai, IN
Package Qubee 256 kbps
Citycell 300 kbps
Druknet 1 Mbps
Airtel 512 kbps
BSNL 512 kbps
Airtel 512 kbps
BSNL 256 kbps
8:00 AM 252.5 271.5 1157.6 1297.8 2366.7 142.8 1072.3
11:00 AM 251.2 267.3 1173.4 1302.1 1591.6 162.9 1052.8
3:00 PM 230.8 177.4 1278.9 1179.4 269.4 163.8 553.4
6:00 PM 251.1 243.6 1061.0 237.8 221.4 268.2 320.6
8:00 PM 256.4 220.6 357.0 147.1 169.1 611.7
11:00 PM 257.8 191.7 1310.4 90.1 175.3 919.5
Average 250.0 228.7 1167.7 947.4 781.0 180.4 755.1
Red = performance < 256 kbps threshold Average is above 256 kbps But below threshold 3/6 9
SLT, Colombo also failed the cut, despite a 2Mbps offering/average of 571 kbps delivery
City, Country
Mumbai, IN New Delhi, IN Colombo, LK
Package Airtel 2 Mbps
MTNL 512 kbps
Airtel 512 kbps
MTNL 2 Mbps
SLT 2 Mbps
Dialog 1 Mbps
Mobitel 1 Mbps
8:00 AM 293.4 375.3 337.6 416.7 1685.8 1082.7 714.9
11:00 AM 316.1 381.3 182.8 425.8 257.2 1054.0 340.5
3:00 PM 304.3 387.8 336.2 343.7 167.4 1054.0 340.5
6:00 PM 275.0 401.6 331.6 437.3 271.1 1401.3 855.0
8:00 PM 322.6 337.2 337.1 429.7 210.1 1355.7 579.9
11:00 PM 308.0 377.1 394.5 429.3 835.1 1398.5 832.4
Average 303.2 376.7 319.9 413.7 571.1 1224.4 610.5
Red = performance < 256 kbps threshold
Access via USB modem (dongle)
10
Sri Lanka’s broadband growth, 2009-March 2013
11
Internet users estimated to be at least 2.5 million
Philippines 2012: Fixed broadband subs 1.47 million (1 m PLDT); Mobile broadband 4 million (2.36m PLDT)?
Broadband in developing Asia appears different
• Mostly wireless, not fixed
• Low prices, but also problems with quality
12
But broadband is unlike voice
• In old style PSTN, most calls are local, few national, fewer international – Local switching possible
• In mobile networks, need to interrogate databases – More signaling traffic going longer distances, but actual
calls and SMS still mostly local
• With broadband (especially in small countries) – Most traffic is international
– Even local websites are hosted abroad
– Even email intended for local persons requires international traffic
13
A chain is as strong as its weakest link Broadband performance is defined by performance over the weakest link
14
IS THE INTERNATIONAL SEGMENT THE WEAKEST LINK?
One cannot focus on local access alone . . .
15
Selected Asian cities: Identifying the weak link (Fixed download)
Source: LIRNEasia, 2011
0
500
1000
1500
2000
2500
800 1100 1500 1800 2000 2300Time
Download from ISP Server (kbps)
Qubee (1 Mbps) Dhaka, BDBanglalion (512 kbps) Dhaka, BDAirtel (256 kbps) Bangalore, INBSNL (256 kbps) Bangalore, INAirtel (2 Mbps) Chennai, INBSNL (256 kbps) Chennai, INAirtel ( 2 Mbps) Mumbai, IN
0
500
1000
1500
2000
2500
800 1100 1500 1800 2000 2300Time
Download from International Server (kbps)
Qubee (1 Mbps) Dhaka, BD Banglalion (512 kbps) Dhaka, BD
Airtel (256 kbps) Bangalore, IN BSNL (256 kbps) Bangalore, IN
Airtel (2 Mbps) Chennai, IN BSNL (256 kbps) Chennai, IN
Airtel ( 2 Mbps) Mumbai, IN MTNL (320 kbps) Mumbai, IN
Airtel (2 Mbps) New Delhi, IN MTNL (512 kbps) New Delhi, IN
SLT (2 Mbps) Colombo, LK Telkom Speed (512 kbps) Jakarta, ID
The problem is international connectivity
16
Selected Asian cities: USB Dongle download (3G networks)
Source: LIRNEasia, 2011
0
200
400
600
800
1000
1200
1400
1600
1800
0800 1100 1500 1800 2000 2300Time
Download from ISP Server (kbps)
Dialog (upto 3.6 Mbps) Colombo, LK
Mobitel (upto 1 Mbps) Colombo, LK
Three (upto 3 Mbps) Jakarta, ID
Globe Tattoo (upto 3.6 Mbps) Manila, PH
0
200
400
600
800
1000
1200
1400
1600
1800
0800 1100 1500 1800 2000 2300Time
Download from International Server (kbps)
Dialog (upto 3.6 Mbps) Colombo, LK
Mobitel (upto 1 Mbps) Colombo, LK
Three (upto 3 Mbps) Jakarta, ID
Globe Tattoo (upto 3.6 Mbps) Manila, PH
The problem is international connectivity, in general
17
Anomalous result in Manila
Selected Asian cities: Latency (RTT) – Fixed: Mostly unsatisfactory
Source: LIRNEasia, 2011
0
50
100
150
200
250
300
350
400
450
0800 1100 1500 1800 2000 2300
RTT
(m
s)
Time
Qubee (1 Mbps) Dhaka, BD Banglalion (512 kbps) Dhaka, BD Airtel (256 kbps) Bangalore, IN
BSNL (256 kbps) Bangalore, IN Airtel (2 Mbps) Chennai, IN BSNL (256 kbps) Chennai, IN
Airtel ( 2 Mbps) Mumbai, IN MTNL (320 kbps) Mumbai, IN Airtel (2 Mbps) New Delhi, IN
MTNL (512 kbps) New Delhi, IN SLT (2 Mbps) Colombo, LK Telkom Speed (512 kbps) Jakarta, ID
300 ms
18
Selected Asian cities: Latency (RTT) – USB Dongle: All above 300 ms
Source: LIRNEasia, 2011
0
100
200
300
400
500
600
700
0800 1100 1500 1800 2000 2300
RTT
(m
s)
Time
Dialog (upto 3.6 Mbps) Colombo, LK Mobitel (upto 1 Mbps) Colombo, LK Three (upto 3 Mbps) Jakarta, ID
Globe Tattoo (upto 3.6 Mbps) Manila, PH Smart (upto 3.6 Mbps) Manila, PH
300 ms
19
Manila
Selected Asian cities: Packet Loss – Fixed (unsatisfactory some of the time)
Source: LIRNEasia, 2011
0
2
4
6
8
10
12
14
16
0800 1100 1500 1800 2000 2300
Pac
ket
loss
(%
)
Time
Qubee (1 Mbps) Dhaka, BD Banglalion (512 kbps) Dhaka, BD Airtel (256 kbps) Bangalore, IN
BSNL (256 kbps) Bangalore, IN Airtel (2 Mbps) Chennai, IN BSNL (256 kbps) Chennai, IN
Airtel ( 2 Mbps) Mumbai, IN MTNL (320 kbps) Mumbai, IN Airtel (2 Mbps) New Delhi, IN
MTNL (512 kbps) New Delhi, IN SLT (2 Mbps) Colombo, LK Telkom Speed (512 kbps) Jakarta, ID
3%
20
Selected Asian cities: Packet Loss – USB Dongle (two unsatisfactory)
Source: LIRNEasia, 2011
0
2
4
6
8
10
12
0800 1100 1500 1800 2000 2300
Pac
ket
loss
(%
)
Time
Dialog (upto 3.6 Mbps) Colombo, LK Mobitel (upto 1 Mbps) Colombo, LK Three (upto 3 Mbps) Jakarta, ID
Globe Tattoo (upto 3.6 Mbps) Manila, PH Smart (upto 3.6 Mbps) Manila, PH
3%
21
Manila
Why?
• Connectivity to the Internet Cloud is a costly input
• Understandably, operators economize on international segment lower performance
– Because overall performance is determined by performance of the weakest link
22
Asian bandwidth is 6x costlier
Source: TeleGeography • Data exclude installation & local
access fees. • GigE = 1,000 Mbps.
$12~$25
$1.75~$4 $1.5~$2
23
What policymakers can do: Asian Highway “Asian Information Superhighway”
• Inadequacy of supply is what causes higher prices – Sole reliance on undersea cables is inappropriate for the word’s largest
continent • Problems exacerbated by frequency of cable cuts, approvals needed for cable-repair
ships to work within territorial waters, high insurance premia driven by piracy in Indian Ocean, etc.
• Take the lead at UNESCAP to promote the laying of cables along the Asian Highway/Trans-Asian Railway to link all of Asia to Europe and the Pacific Rim – Must be an OPEN ACCESS mesh network, ideally owned by entity other than
incumbent telecom operator in each country learning from problems experienced in ADB-funded SASEC, for example
• Necessary solution for land-locked countries • Terrestrial cables will COMPLEMENT submarine cable & bring down prices
for all, including those countries that do not use terrestrial cables • Encourage more PPPs in undersea cables, especially in underserved areas
such as the Pacific and Bay of Bengal
24
What policymakers can do: Asian Highway “Asian Information Highway”
• Increased supply of backhaul will bring down wholesale prices for all, island nations as well as landlocked countries
• But these reductions will be reflected in retail prices only if domestic policy allows it – A terrestrial network along highway traces will provide both
international and terrestrial backhaul • Publish domestic leased-line prices using standard definitions such as
2MB/2km; 2MB/100km • Few standard measures exist for international leased lines, but more
transparency is needed here as well
– Give “build or buy” option to all users • If there are economic advantages to a single supplier, there is no
reason to have legal barriers to others doing “foolish” things
– Ensure competition exists in supply of broadband, wireline & wireless
25
What of satellites?
• Latency of 300 ms required for most applications
– Singapore IDA standard, now adopted by others
• Not favored by BPO industry
• Generally useful as backup/redundancy
26
National and regional Internet exchanges
• More traffic is bypassing North America, but given availability of capacity and prices, different regions within Asia will have to decide who to exchange with
27
ACCESS NETWORK (THE LAST MILE) The next link of the chain . . .
28
Fiber or wireless?
• Some are passionate about Fiber-to-the-Home (FTTH), arguing that anything less is second-class
• Others see no alternative to wireless access, given
– Purchasing power of consumers in most countries
– Cost of laying wire (copper or fiber) to homes (in much of Asia there are no wireguides that can be upgraded)
29
Middle ground
• Everyone, rich and poor, will access the Internet wirelessly – Those in developed market economies and those with wealth living in
densely populated cities of developing countries, over a few meters – The rest of us, over a few kilometers
• Fiber should be pushed out as far as possible (India’s current thinking on utilizing the Universal Service Fund is an example)
• Access should be open • As many operators as possible should be permitted to link to
the fiber, using multiple technologies, wire and wireless – If wire is superior in meeting people’s needs, it will triumph over time
• Government intervention should be – Technology neutral – Focused on ensuring that all settlements have fiber within a
reasonable distance – Access is open to all operators
30
New Zealand as a possible model to emulate
• Vertical disintegration of incumbent telco is the optimal solution to ensure open access
• Second and third best solutions need to be worked out if NZ solution is not feasible
• Removal of barriers to entry at the access level, subject to
– Release of frequencies
– Reasonable access to rights of way
31
Broadband without electricity?
• Can have broadband in areas without electricity
– But it’s costly
• Much of the technical innovation needed to get broadband to people is in energy
32
Pakistan has made renewables a condition of universal service subsidies
The economics of renewable energy look much better in relation to expensive and erratic conventional energy
33
Energy for terminal devices
• Less of a challenge than energy for network equipment
– People develop workarounds
• Can provide recharging opportunities from base stations that are powered by renewable energy
34
PRICE AND ATTRACTIVE APPLICATIONS
The critical links of the chain: what matters to the consumer
35
0
10
20
30
40
50
60
Ban
glad
esh
Sri L
anka
Ch
ina
Pak
ista
n
Ind
ia
Uzb
eki
stan
Ke
nya
Egyp
t
Vie
tnam
Sud
an
Iran
Eth
iop
ia
Cam
bo
dia
Thai
lan
d
Gh
ana
Uga
nd
a
Tan
zan
ia
Hai
ti
Ind
on
esi
a
Alg
eria
Ph
ilip
pin
es
Tun
isia
Bo
livia
Gu
ate
mal
a
Mo
zam
biq
ue
Nig
eri
a
Sen
egal
AV
ERA
GE
Syri
a
Ho
nd
ura
s
Cô
te d
'Ivo
ire
Kaz
akh
stan
Ecu
ado
r
Do
min
ican
Rep
ub
lic
Gu
ine
a
Sou
th A
fric
a
Mad
agas
car
An
gola
Zim
bab
we
Bu
rkin
a Fa
so
DR
C
Co
lom
bia
Zam
bia
Mal
awi
Ch
ile
Cam
ero
on
Mo
rocc
o
Turk
ey
Ch
ad
Arg
en
tin
a
Pe
ru
Bra
zil
USD
pe
r m
on
th
Monthly TCO(USD)
Internet premium(USD)
Voice + SMS TCO: Brazil = Bangladesh x 23 Voice, SMS & Internet TCO: Morocco = Sri Lanka x 57 Different business model in sub USD 10 countries? Budget Telecom Network (BTN) model
Ave: USD 11.47
Ave with Internet premium: USD 15.05
Source: Nokia
Nokia total cost of ownership study 2011
36
Attractive content: What governments can do
• Asian prices as measured by Nokia are reasonable (only Philippines is above average, but even here cause may be voice prices in the basket)
• Ensure conditions for supply of attractive applications are maintained – Do not follow the siren call of “access charges” or sending-party-
network-pays principle still being promoted by certain parties
– Incubators, meet ups, other ways to promote innovation in the mobile apps space
– Open up government data so mobile app developers have raw material to work with
37
Chain is a useful metaphor. But we’re really dealing with networks. A resilient mesh made up of multiple
chains . . .
38