Transcript VoIP-Roundtable - Columbia University

911 services: wireline,
wireless and VoIP
Prof. Henning Schulzrinne
Dept. of Computer Science
Columbia University, New York
[email protected]
VoIP Roundtable, Washington, DC
June 24, 2004
Overview






E911 for wireline
E911 for wireless (Phase II)
Differences between phone system and
VoIP
Requirements and opportunities
Internet standardization efforts
Funding and regulation challenges
Components of emergency calling
Three core components that need to be
replicated – everything else are
implementation details

1.
2.
identifying emergency calls (“911”)
determining the right emergency call center
(PSAP) for current caller location


3.
coarse-grained location
currently, two main databases: ALI for number-tolocation mapping and MSAG for address verification
deliver caller location to PSAP

fine-grained location
E911 for wireline
CAMA or SS7
555-1234
313 Main St
CO Switch
LEC
network
PSAP #1
CAMA or PRI
delivers ANI
(555-1234)
Tandem Switch
(911 Selective Router)
555-1234
 PSAP #1
MSAG
ANI: 555-1234
 313 Main
ALI
100-500 Main Street  ESN 1789
private
data link
555-1234  PSAP #1, 313 Main St
provisioned
updates
verify address
validity
Wireless 911

Phase I (April 1998)




Route all call to the
appropriate PSAP based
on call sector
Provide cell/sector
location data to PSAP
Provide call back number
to PSAP
Phase II (October 2001)

Phase I + latitude and
longitude
67%
95%
handset
50m
150m
network
100m
300m
Wireless 911: Phase 2
MSC
pANI
(ESRD or ESRK)
ISUP
r s
i l
t
r
i
y N et
LEC
selective router
Wireless Tower
PDE
MPC/SCP
E2
S
A-GPS, UTDOA, …
t
w .a p
dynamic
updates
.c o
ESRK = unique for call
ESRD = unique for location
ALI
ESRK or ESRD 
coordinates
callback number
Problems with existing 911
system

1970s technology:



CAMA trunks induce long call setup delays
limited in ability to transfer information (8-10 digits)
sometimes, 2,400 baud modems for database access


gets complicated if multiple providers




increases call setup delays
ILEC vs. CLEC
multiple wireless providers in one state
tied to ILEC rate centers and other PSTN routing
artifacts
hard to move PSAPs on short notice (e.g., emergency
evacuation)

can’t just plug into any network termination
PSTN vs. Internet Telephony
PSTN:
Signaling & Media
Signaling & Media
Internet
telephony:
China
Signaling
Signaling
Media
Belgian customer,
currently visiting US
Australia
How does VoIP differ from
landline and wireless PSTN?

Telephone companies are no
longer needed





new location, but same identifier
there are still carriers for DSL and
cable “IP dial tone”
but unaware of type of data carried
(voice, web, IM, …)
VSP may be in another state or
country
anybody can be their own “VSP”
Corporations and universities
don’t have email carriers, either
Yahoo

voice service provider
[VSP]
MCI
All devices are nomadic
ISP
(IP, DHCP)
NYSERNET

dark fiber
provider
(λ)
(TCP, RTP, SIP)
The role of phone numbers and
identifiers



Wireline  line, device, subscriber & location
Wireless  device, but not location
VoIP (phone number and URIs):

mostly identifies person, not device






multiple devices located in different states can share the
same number
however, may not have a phone number
if it does, area code may be from different state
than customer billing address
multiple devices
device can move, while number stays the same
not related to ISP
Why is VoIP ≠ wireless?

VoIP devices may not have phone numbers as lookup keys


e.g., sip:[email protected]
Location information for devices is civil, not
longitude/latitude


e.g., service address for VSPs
GPS not available (nor functional) on indoor devices



Cell phones don’t work in our building…


plus, accuracy of 50 m (67%) or 150 m spans many buildings…
no floor information
so A-GPS is unlikely to work there, either
Plus, wireless E911 complexity due to old signaling
mechanism


expensive and complicated to connect to multiple wireless
operators
proposals to use IP-based solutions
50m
Objectives for IP-based 911

International







re-use standard protocols (SIP,
DNS, DHCP, HTTP, XML, …)
avoid repeat of CAMA trunks
Resilient

easily re-route calls to any
number of backup PSAPs
integrity, privacy and
confidentiality, protection
against denial-of-service attacks
Technology-independent


users can test operation
without tying up operator
resources
Secure

COTS (commercial off-the-shelf)


integrate alternate modalities
such as text (TDD) and video
(sign language)
Testable

Multimedia


devices must work anywhere
independent of local emergency
number
international roaming

do not depend on (e.g.,)
specific wireless or link
technology
Pro-competitive

does not require carriers or
gatekeepers
Opportunities for I911

More robust




multiple networks and
interfaces increase disaster
resiliency
operations can be moved easily
to any network-connected
location



Additional services



text chat replacing TTY
video and images for
situational awareness and
instructions to civilians at
scene of accident
additional data


integration with telematics
providers
general awareness of call
volume, origin and type
information flow back to PSAP

multimedia

Better integration with first
responders and public safety



see 9/11 evacuation
hand off call data, not just
remote printing of address
alerts and notifications to public
safety and the public
Cheaper to build and operate

hazmat data
accident data (impact velocity,
airbags, fuel spill, …)

currently, small niche 
expensive equipment,
specialized circuits, slow
upgrades
should be able to leverage
almost existing technology 
lower risk
Three stages to VoIP 911: I1 
I2  I3

I1:



I1 & I2:




I3:



may use administrative line to deliver call
no location delivery
no modification to PSAP
calls delivered to PSAP via existing technology
delivers location information
deliver calls to PSAP via VoIP
including circuit-switched calls
I1, I2 and I3 will likely co-exist for some time

design to allow local upgrades, without national or statewide coordination
Three stages to VoIP 911
when
deployable
?
use 10digit
admin.
number?
mobility
caller
location
conveyed
to PSAP?
PSAP
modification
ALI (DB)
modification
new services
I1
now
allowed
stationary
no
no
no
none
I2
December
2004
no
stationary
nomadic
yes
no (10-digit)
yes
none
I3
specified
by late
2004
no
stationary
nomadic
mobile
yes
IP-enabled
replaced by
DNS
global number
portability
multimedia
international
calls
IETF and NENA I3
standardization efforts

IETF = Internet Engineering Task Force =
international open standardization body
“911”
sip:sos@
include
civil and/or geo
911  sos
112  sos
sip:[email protected]
provide location
(civil or geo)
DHCP
cn=us, a1=nj, a2=bergen
Regulatory challenges

Uniform technology


but avoid things that don’t work internationally
Distributed responsibility:


VSP does not know location
Residential user may have any number of VSPs




like prepaid calling cards
and may use any one of them for calling 911
ISP does not know VSP and whether call is voice
or not
 ISP needs to provide location information to
end user  regulatory mandate!
Funding challenges

Existing line/number model
does not work




number assigned may not
match location of customer


VoIP has built-in global number
portability
can’t require to keep track of
6,000 different county 911 fees
and recipients
Funding collector should

longer term, no more “lines”
and numbers
every person may have multiple
identifiers
Short-term vs. long-term
options
Even small VSPs have a national
footprint






have direct customer
relationship
know accurately where
customer lives
have regional footprint
have modest collection and
distribution costs
Some options:




facilities-based (broadband) ISP
local & state taxes
non-telecom utilities (water,
gas, electric) that reflect
residency
collect directly from household

like home owners insurance
Conclusion

Existing 911 system closely tied to phone system
history





VoIP offers opportunity to increase robustness, offer
new services and decrease costs
Initial international and US standardization efforts in
progress


number as universal identifier
close affiliation with telephone switches
incremental, constrained evolution
IETF and NENA collaboration combines 911 and Internet
expertise
Initial prototypes and demonstration systems in
development
Additional I1 and I2 information
Example I1 solution #1
SS7
CLEC #1
Example: VoxPath
End Office
Switch
LAN
IP Phone
Customer
Signaling
Gateway
PSAP
Softswitch
Broadband
Network
Media
Gateway
End Office
Switch
Selective
Router
1
2
3
4
5
6
7
8
9
*
8
#
Call Taker
POP
CLEC #2
Mark Lewis
I1 Solution #2
Level3
Customer
Softswitch
Dedicated
911 trunks
PSTN
E911
Tandem
End
Users
Public Internet
or
Private IP
Network
E911
Tandem
IP Phones
ALI
DB
PBX
(emergency lines)
Level 3 911
Softswitch
Network
ACD
PSAP
Mark Lewis
Possible I2 architecture
IP
INVITE sos
PSTN
MG
Selective
Router
ISUP
CAMA
ESRK, DN  loc
ALI
SIP
PUBLISH
ALIFE
E2+
ESP
PAM
Local
ALI
National
Voice N/Ws
Emergency Services N/W
based on slide
by Martin Dawson