quality of service regulation -...

Quality of Service Regulation

Helani Galpaya Taungoo, Myanmar

28 Sep – 2 Oct 2013

WHAT, WHY & HOW

Trade off: price vs. quality

• A consumer buys a price-quality bundle – High quality – high price bundles: luxury travel, first

class air tickets

– Low cost – Low quality bundle: travel on budget airlines

• Budget telecom network model: – Very low cost, problematic (low) quality

• ‘Quality costs money’

Quality is important to consumers

• Ability to get dial-tone

• Ability to complete a conversation without the call dropping

• Ability to move around with a mobile phone, and complete a conversation, without ‘dead spots’

• Ability to download a song in ‘reasonable’ time

• Etc…

• Impacts take-up to an extent

• But so does price

Quality is important to operators

• An operator intentionally misroutes calls from another operator

• Incumbent operator who runs internet (data) exchange ‘chokes’ the data connections from competitors?

• Etc.

Concepts of quality extend beyond call quality and data speeds

• Service provision – Waiting list for a service

• Faults – Call drops/data connection unavailable; voice/data

quality

• Fault clearance – How long it takes to correct faults in service when

reported

• Billing – Accuracy, corrections when reported, refunds

ITU has limited list for voice quality indicators Trying to develop data quality indicators

How to monitor quality: traditionally

• Operators have most/all information about performance

• Reports to regulator

• Regulator analyzes

Changes to quality monitoring approaches

• Involve users – Crowdsourcing

– Example in Broadband, 2nd half of this talk

• Monitored not just by the regulator – Other interested parties: e.g. civil society

• Also a change of indicators from hard soft – Hard: data on actual network performance

– Soft: data on user satisfaction

Quality impacts the user-experience and may impact adoption of a service by users

• Some consumers may not take up a service if quality is low – But others may be willing to put up with bad

quality for lower prices

• ‘Poor quality, but affordable’ is better than ‘high quality but unaffordable’

• Early days: let operators roll out networks services; let consumers get connected

Regulator does not punish low quality (only monitors). But others can

• In a market with comparable mobile prices – If customers know operator A gives less call drops, they

will choose A and not B – If customers knows B gives better data download

speeds than A, people who value data will choice SIMs from B

• Normally, only regulator knows performance data • But regulator should strategically release this data

– So media can write about it – Operators (the ‘good performers’) start using it in their

advertising • When there is CHOICE, people ‘vote with their

feet’

In the long run, increase competition (i.e. get more operators)

• With increasing competition prices keep coming down penetration increases

• Early days, operators focus on gaining market share – Just being there is enough to get new customers

• But as market saturates, operators need to differentiate themselves – To get customers to move from network A to B,

based on better quality

• Creates incentives for increasing quality

MEASURING BROADBAND QUALITY A concrete example from LIRNEasia’s research/work

What to measure? Think beyond download speed

+++ Highly relevant; ++ Very relevant; + Relevant; - Irrelevant

Service Download (kbps)

Upload (kbps)

Latency (Round Trip Time, RTT) (ms)

Jitter (ms)

Packet Loss (%)

Browsing (Text) ++ - ++ - -

Browsing (Media) +++ - ++ + +

Downloading +++ - - - -

Transactions - - ++ + -

Streaming media +++ - ++ ++ ++

VOIP + + +++ +++ +++

Games + + +++ ++ ++

- RTT has implications on client-server interactive systems

- Jitter adds to the ‘noise’ of the transmission

- Packet Loss affects streaming media

Source: Gonsalves, T.A & Bharadwaj, A. (2009) 14

User’s

ISP

InternetInternet

User

1st Entry point to US

Yahoo.com

ISP

National Server

3 Domains…

Servers of

the User’s

ISP

Server of another

operator, in the

country

First entry point to the

cloud (US/UK/China/…..).

Outside country

WHERE the data is hosted

makes a difference.

We focused on 3 domains

Software based, new testing methodology

• Developed methodology and software in partnership with Indian Institute of Technology (IIT), Madras

• Testing in multiple domains

- ISP, National and International (yahoo)

• Multiple broadband packages or links

– 256kbps – 2Mbps

• Multiple times a day

– 6 times a day; ranging from peak to off peak

• Multiple Locations (cities)

Quality testing can be done by different parties

17

Diagnostics conducted by

Pros Cons

Service Providers (ISPs/operators)

• Easy to implement

• Results based on equipment placed in the most optimized points of the network • Not representative of the ‘actual’ speeds

Users • Represents actual user experiences

• Assumes user’s PC is virus-free, with no parallel process running etc. • Dependant on user’s willingness to participate • Large files impact user’s data limits

National Regulatory Authorities (NRAs)

• In a position to request operator involvement when necessary

• Known test locations will prompt operators to optimize networks in selected areas

We tried different approaches

• First, crowdsourcing – Blogger and telecenter operator meet-ups+ training;

– Small community; High transaction cost abandoned after 1 or 2 rounds

– But worked in Africa: different incentives?

• Then (and now) test it ourselves – researchers spread across Asia (cover their cost)

– Successful; but limited to where we can find a tester

• Given free to regulators. Got them to adopt – ICTA (Mauritius), BICMA (Bhutan), CAM (Maldives),

Fiji and Pacific Islands

SOME RESULTS

Actual performance differs from advertised. In surprising ways

0

100

200

300

400

500

600

700

800

900

1000

08:00 11:00 15:00 18:00 20:00 23:00

BSNL (512 kbps) Bangalore, IN BSNL (256 kbps) Chennai, IN

SLT (2 Mbps) Colombo, LK 3BB (5 Mbps) Bangkok, TH

3BB (5 Mbps) Chiang Mai , TH

0

250

500

750

1000

0800 1100 1500 1800 2000 2300 Time

Druknet Leased line (512 kbps) Thimphu Tashi InfoComm Leased line (256 kbps) Thimphu Druknet (prepaid) ADSL (256 kbps) Thimphu Druknet (prepaid) ADSL (256 kbps) Paro Bhutan Telecom 3G (550kbps to 1Mbps) Thimphu

Bhutan: Delivered (actual) as % of

advertised, from international server

India, Sri Lanka, Thailand: Delivered

(actual) as % of advertised, from local

server

Ra

tio

, a

ctu

al to

ad

ve

rtis

ed

(%)

Ra

tio

, a

ctu

al to

ad

ve

rtis

ed

(%)

Source: LIRNEasia, 2009

South Asian users get low value for money compared to N American counterparts

Do

wn

load

sp

eed

per

Do

llar

(kb

ps p

er

US

D)

0

10

20

30

40

50

60

70

80

90

100

0800 1100 1500 1800 2000 2300

Package per month in

USD

Sirius (256 kbps) Dhaka, BD 15

SKYbd (256 kbps) Dhaka, BD 15

BSNL (256 kbps) Bangalore, IN 16

Airtel, (256 kbps) Delhi, IN 16

SLT (2 Mbps) Colombo, LK 48

Dialog (2 Mbps) Colombo, LK 44

Bell (6 Mbps) Ottawa, CA 60

Rogers (10 Mbps) Ottawa, CA 45

Verizon (3 Mbps) Buffalo, US 30

Comcast (6 Mbps) Denver, US 60

All high values for kbps/USD from N America

Indian users get lower speeds when downloading international content

0

1000

2000

3000

4000

5000

800 1100 1500 1800 2000 2300

Do

wn

load

sp

eed

(kb

ps)

Time

India - Download from ISP server

Airtel (512 kbps) Bangalore, IN BSNL (512 kbps) Bangalore, IN

Airtel (512 kbps) Chennai, IN BSNL (256 kbps) Chennai, IN

Airtel ( 2 Mbps) Mumbai, IN MTNL (512 kbps) Mumbai, IN

Airtel (512 kbps) New Delhi, IN MTNL (2 Mbps) New Delhi, IN

0

1000

2000

3000

4000

5000

800 1100 1500 1800 2000 2300

Do

wn

load

sp

eed

(kb

ps)

Time

India - Download from International server





…as do Sri Lankan users

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

800 1100 1500 1800 2000 2300

Do

wn

load

sp

ee

d (

kbp

s)

Time

Sri Lanka - Download from ISP server

SLT (2 Mbps) Colombo, LK

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

800 1100 1500 1800 2000 2300

Do

wn

load

sp

ee

d (

kbp

s)

Time

Sri Lanka - Download from International server


Round trip time deteriorates (increases) also when going international from India ..

0

100

200

300

400

500

600

700

800

0800 1100 1500 1800 2000 2300

RT

T (m

s)

Time

India - RTT when pinged to ISP server





0

200

400

600

800

0800 1100 1500 1800 2000 2300

RTT

(m

s)

Time

India - RTT when pinged to International server





….and from Sri Lanka

0

100

200

300

400

500

600

700

800

0800 1100 1500 1800 2000 2300

RT

T (m

s)

Time

Sri Lanka - RTT when pinged to ISP server


0

100

200

300

400

500

600

700

800

0800 1100 1500 1800 2000 2300

RT

T (m

s)

Time

Sri Lanka - RTT when pinged to International server


SO WHAT? DID ANYTHING CHANE?

Published ads in newspaper during SATRC meetings

• In the newspaper delivered in conference hotel

• Followed by other publicity about advertising ‘untruths’ of operators

– Delivering well beyond advertised

– Advertising theoretical maximum speeds, not actually possible

At least one operator stops advertising theoretical speeds. Within 4 months, TRCSL starts its own testing.

Source: http://www.trc.gov.lk/broadband/broadband-initiative.html

http://www.trc.gov.lk/broadband/broadband-initiative.html



THANK YOU

quality of service regulation -...

Documents