Transcript CAT Forum -- October 12, 2004

Re-inventing the
Telephone System:
The Third Generation
Henning Schulzrinne
Dept. of Computer Science
Columbia University
CAT Forum -- October 12, 2004
Overview







1st generation: analog
2nd generation: digital circuit switched
3rd generation: packet-switched
What is VoIP? Why now?
A short history
Wireless VoIP
Context-aware communications
VoIP  IM, presence
Challenges ahead
Lifecycle of technologies
traditional technology propagation:
military
opex/capex
doesn’t
matter;
expert
support
Can it be done?
corporate
capex/opex
sensitive,
but
amortized;
expert
support
Can I afford it?
consumer
capex
sensitive;
amateur
Can my mother use it?
Internet and networks
timeline
theory
1960
university
prototypes
1970
port
speeds
Internet
protocols
production use
in research
1980
100 kb/s
email
ftp
queuing
architecture
commercial
early residential
1990
1 Mb/s
DNS
RIP
UDP
TCP
SMTP
SNMP
finger
routing
cong. control
broadband
home
2000
10 Mb/s
2010
100 Mb/s
ATM
BGP, OSPF
Mbone
IPsec
HTTP
HTML
RTP
DQDB, ATM
QoS
VoD
1 Gb/s
XML
OWL
SIP
Jabber
p2p
ad-hoc
sensor
Earlier PSTN changes

starting in 1980s:




end systems relatively unaffected
few additional services



analog  digital transmission
in-band  out-of-band (SS7) signaling
800#
CLASS services (caller ID, call waiting)
customer relationship largely unaffected

except CLECs and reselling
100
Technology evolution of the
PSTN
90
80
70
60
electromech
analog
digital
50
40
30
20
10
0
1980 1985 1987 1990 1995 2000 2001
SS7: 1987-1997
What is VoIP?



Voice-over-IP = Internet telephony
 “Internet telephony refers to communications
services—voice, facsimile, and/or voice-messaging
applications—that are transported via the Internet,
rather than the public switched telephone network
(PSTN). The basic steps involved in originating an
Internet telephone call are conversion of the analog
voice signal to digital format and
compression/translation of the signal into Internet
protocol (IP) packets for transmission over the
Internet; the process is reversed at the receiving end.”
(IEC)
Not a single technology, but combination of Internet
technologies
Now typically voice only, but easily extended to video
Brief history of packetized
voice












1969: ARPAnet, predecessor of modern Internet
1974: real-time packetized voice (early Internet)
1990: primitive version used for transatlantic calls (G.764)
1991: DARTnet (test network) audio experiments using Sun
workstations
1992: first IETF multicast audiocast
1992: RTP (transport) draft
1995: first commercial PC-to-PC voice software (Vocaltec)
1995/1996: first PC-to-PC and PC-to-phone services
(Net2Phone, DialPad, Vocaltec, …)
1996: first version of SIP and H.323 standards
~2000: first service providers
~2002: first large-scale consumer services
2002: 3G wireless specifies Internet multimedia subsystem
How has the industry progressed




Softswitch networks carry approximately 2 billion
minutes/day vs. 2.3 million in 1999
Services, such as IP-Centrex, are quickly being
adopted by enterprises
10% of all international voice traffic to/from U.S.
carried on IP
 primarily prepaid calling cards
IP-enabled handset sales over 4.5 million units in
2002
 35% of all total premise sales are IP – enabled
 IP handset costs drop from $600 in 2001 to $99
today
TIA, 2003 Telecommunications Market Review and Forecast
Gartner Dataquest, 2001-2007, United States: Fixed Public Network Services
Gartner Dataquest, 2002 Premises Based Equipment Sales
Jack Waters
Fall VON 2003
VoIP penetration
14
residential & small
business
12
10
8
Cable
Non-facilities
%
6
4
2
0
YE2003 YE2004 YE2005 YE2006 YE2007 YE2008
Glen Campbell
Telecom & Cable Analyst
Merrill Lynch Canada
May 2004 (CITI VoIP workshop)
Vonage Subscriber Growth
subscribers in thousands
250.0
200.0
150.0
100.0
50.0
4
A
ug
-0
04
)
(E
3
t-0
Fe
b-
3
Se
p
A
ug
-0
3
Ju
l- 0
3
03
Ju
n-
ay
-0
3
M
A
pr
-0
ar
-0
3
M
03
Fe
b-
3
n0
Ja
-0
2
D
ec
-0
2
N
ov
2
ct
-0
O
-0
2
Se
p
A
ug
-0
2
-
*Cable Datacom News Round Up, September 1, 2003
Why has it taken so long?


VoIP technology development since 1995
Web: worked on dial-up, motivated broadband



deployment from 1992 to 2000
VoIP: not usable on dial-up, spurred by residential
broadband
More than just protocols needed:




eco-system (management, configuration, OSS, …)
interoperability
spectrum of products – low to high end
interoperation with legacy equipment
Technology introductions
Source: OECD, 2003
Conditions for VoIP

Multimedia PC with low-latency OS


earlier Windows versions not suitable
Broadband access for residence

modem adds significant delay
High-speed switched LANs for
businesses
  only feasible since mid-1990’s

Total high-speed lines
FCC, 2004
DSL usage
DSL Forum, Sept. 2004
DSL penetration
DSL Forum, Sept. 2004
Who provides VoIP service?
service +
gateways
voice service
provider
(Vonage, Lingo, Packet8)
service +
gateways +
IP network
service +
gateways +
IP network +
access
cable
providers
long-distance
carrier
(e.g., AT&T, MCI)
DSL
(ILEC)
Motivations for VoIP
financial
services
security
• better voice quality
• access fee
• taxes
• monopoly rents
• local-loop access
• separate wiring plant
• cheaper services
(caller ID, etc.)
• higher network
efficiency
possible
• user-defined
services
• video and app.
sharing
• integration of
presence
• abundance of
identifiers
• mobility
• media encryption
• signaling encryption
• user authentication
VoIP models: PBX
analog
telephone
adapter
call server
proxy server
“softswitch”
VoIP gateway
LAN
enterprise
IP
VoIP models: IP Centrex
service
provider
IP
LAN
enterprise or residence
Some differences: VoIP vs.
PSTN




any media quality
(e.g., TV and
radio circuits)
interactive games
No need for telephone
company
Yahoo


voice service provider
MCI

Separate signaling
from media data path
But, unlike SS7, same
network  lower call
setup delay
Avoid CTI complexity
of "remote control"
Mobile and wireline
very similar
Any media as session:
ISP
(IP, DHCP, DNS)
NYSERNET

(RTP, SIP)
dark fiber
provider
VoIP components

Re-uses whole Internet protocol architecture and transmission
infrastructure



IP, UDP for transport
TLS and S/MIME for security
HTTP for configuration

signaling
transport
directories
ENUM
H.350
provide
URI
SIP/SDP
H.248
MGCP
H.323
provide
destination
address
RTP
codecs
(G.7xx,
H.26x)
SIP trapezoid
outbound proxy
destination proxy
(identified by SIP URI domain)
1st request
SIP trapezoid
2nd, 3rd, … request
[email protected]:
128.59.16.1
registrar
voice traffic
RTP
Example SIP phones
about $85
PSTN vs. Internet Telephony
PSTN:
Signaling & Media
Internet
telephony:
Signaling & Media
China
Signaling
Signaling
Media
Belgian customer,
currently visiting US
Australia
SIP as service enabler

Rendezvous protocol


lets users find each other
by only knowing a
permanent identifier
Mobility enabler:

personal mobility
•

terminal mobility
•

one terminal, multiple IP
addresses
session mobility
•

one person, multiple
terminals
one user, multiple
terminals in sequence or
in parallel
service mobility
•
services move with user
Changes caused by VoIP

Access independence: single-function
network to voice-over-any-network

separation of transport and services
Transition from “polling” service (try
until user happens to be available) to
“presence” service
 Voice special  voice just one media
among many

(Early) Adulthood

“fully developed and mature”
Not quite yet, but no longer a teenager
 probably need another 6 years to be
grown up…


Responsibilities:
Dealing with elderly relatives  POTS
 Financial issues  payments,
RADIUS
 Family emergencies  911

Emerging technologies

Core VoIP technology largely finished





Presence  from “polling” to “status report”




deployment largely due to cost savings, not new
services
toll and fee bypass
integrated infrastructure (LAN & WAN)
extend “PBX” reach to home and branch offices
special case of event notification
events as common infrastructure for services
location-based services
Integration of IM and VoIP


often used in same conference (side channel)
IM as initiator of real-time voice/video
Near future: Location-based
services

Finding services based on location




physical services (stores, restaurants, ATMs,
…)
electronic services (media I/O, printer, display,
…)
not covered here
Using location to improve (network) services

communication
•

configuration
•

devices in room adapt to their current users
awareness
•

incoming communications changes based on where I am
others are (selectively) made aware of my location
security
•
proximity grants temporary access to local resources
Location-based IM &
presence
User service creation



Tailor a shared infrastructure to
individual users
traditionally, only vendors (and
sometimes carriers)
learn from web models
 not one “killer application”
 grass-roots applications not
foreseen by carriers
programmer,
carrier
end user
network
servers
SIP servlets,
sip-cgi
CPL
end system
VoiceXML
VoiceXML (voice),
LESS
Near future: Multimedia



Wideband audio
 “better than phone
quality”  lectures,
discussions, speaker
phone
 better codecs  same
bandwidth as existing
NB codecs
Video phone itself remains
niche application
 given low incremental
cost, may be viable
 useful for sign language
Video for group meetings
 capture whiteboard



Shared applications
(WebEx, etc.)
 still requires
standardization
Instant messaging
 side channel
Better means of
coordination (floor
control)
wideband audio
Near future: VoIP over WiFi

Not fundamentally different from landline
VoIP





combination cellular + WiFi = wide-area +
“cordless” phone
Small packet sizes make VoIP over WiFi far
less efficient than nominal data rate
Hand-off delay between different base
stations  interruptions  CU modified
hand-off algorithm
Delay jitter with high loads  new
scheduling algorithms
L3 hand-off across different network types
Challenge: Global
interconnect

Currently, each VoIP “network”
largely isolated



Long-term solution: ENUM DNS
listing


interconnect via PSTN even if
both endpoints are on IP
interconnect via few peering
points even if neighbors
administration appears difficult
Short-term for pure-IP (FWD,
etc.): special number prefixes
GW
VSP A
GW
Enterprise B
Challenge: CALEA (lawful
intercept)


Existing models assume congruence of signaling
and voice flows
Challenges:





voice service providers outside the US
signaling-only providers or no voice providers
end-to-end media and signaling encryption (Skype,
SRTP)
 Intercept IP traffic, not application
Assume that long-term, all application traffic (except
browsing of public web pages) will have strong
encryption
Challenge: User-programmable
and context-aware services

Universal reachability  control reachability
in time and space by context


allow callee to decide reachability (defer and
decline communication)
choose appropriate media (text, automated
data response)
time
Call Processing Language (CPL), sip-cgi, …
capabilities
caller preferences
location
location-based call routing
location events
activity/availability
presence
sensor data (mood, bio) not yet, but similar to location data
Challenge: Spam prevention

Currently, telemarketing restricted to in-country
calling




With VoIP, few economical constraints on automated
calls from anywhere
Also, SPIM (instant message spam)
Cannot use content-based filtering
Public key infrastructure (PKI) for individual
verification has never scaled



provide domain-level verification (~ TLS) in signaling
blacklists and whitelists
• may depend on local domain policies for user
verification
reputation-based systems
Challenge: Service reliability

“QoS”  service availability





loss of network connection
loss of infrastructure components
• DNS, SIP servers, DHCP, …
bursts of packet loss  cannot be repaired at end
system
sustained high packet loss (> 10-15%)
Current service availability probably around 99.5%

realistic goal: 99.9% (10h/year) to 99.99% (1h/year)
Challenge: Emergency
calling

911 calling system largely unchanged since 1980s



Fundamental differences for VoIP:





call routing to appropriate destination
deliver caller location information
may not have phone number
may be no “phone company”
identifier does not describe location
location determination more difficult
Also use solution for “311” and other location-based
call routing systems
Three stages to VoIP 911
spec.
available?
use 10digit
admin.
number?
mobility
callback
number
to
PSAP?
caller
location
to
PSAP?
PSAP
modification
ALI (DB)
modification
new services
I1
now
allowed
stationary
no
no
no
no
none
I2
Dec. 2004
no
stationary
nomadic
yes
yes
no (8 or 10
digit)
update
none
I3
late 2004
no
stationary
nomadic
mobile
yes
yes
IP-enabled
ALI not
needed
MSAG
replaced by
DNS
location inband
GNP
multimedia
international
calls
Prototype
* gray features in progress.
Call taker setup
SIPc client receives calls
GeoLynx software displays
caller location
Conclusion

VoIP on cusp of widespread deployment:






Focus may shift from “bare-bones” VoIP to
context-aware communications
Operational and technical challenges


commercial-grade VoIP products
mature standards for key components
widespread broadband availability
better Internet QoS
911, CALEA, network reliability, user-defined
services, multimedia
Thus, roughly where PSTN was in 1980 