IPv6 And Cellular Telephony

1
IPv6 And Cellular Telephony

• Phil Karn
• VP Technology
• Qualcomm, Inc
• 26 June 2003

2
Safe Harbor (QUALCOMM Disclaimer)

• Before we proceed with our presentation, we would
  like to point out that the following discussion
  will contain forward-looking statements from
  industry consultants, QUALCOMM, and others
  regarding potential market size, market shares,
  and other factors which inherently involve risks
  and uncertainties.
• These and other risks and uncertainties relating
  to QUALCOMMs business are outlined in detail in
  our most recent 10-Q and 10-K forms filed with
  the Securities and Exchange Commission.
• Please consult those documents for a more
  complete understanding of these risks and
  uncertainties.

3
My Disclaimer

• This talk includes my personal opinions.
• I am not speaking for Qualcomm.
• Qualcomm may or may not agree with me.
• (But they should.)

4
The State of Cellular Telephony

• A plethora of incompatible physical layers
• AMPS (analog)
• IS-54 TDMA
• GSM
• CDMA (IS-95/IS-2000, 1x, 1xRTT, 1xEVDO, etc)
• WCDMA
• even 802.11 b/a/g/etc
• Cheap, mass produced phones
• Expensive, inflexible base station equipment

5
Interoperability

• There's an increasing realization of the need of
  cellular users to roam between dissimilar
  cellular networks
• Administrative barriers
• Technical barriers
• Users don't want to be bothered by the details

6
A Brief Internet History

• ARPANET started in 1969
• Ethernet in mid 1970s
• DARPA Packet Radio in 1970s
• The need quickly appeared to join these
  dissimilar networks in a uniform way
• TCP/IP and the Internet were the result
• Sound familiar?

7

• The solution to the cellular interoperability
  problem is IP!

8
Other Useful Properties of IP

• End-to-end architecture
• recently rediscovered and hyped as "peer to peer"
• place application-specific mechanisms at end
  points
• keep network simple and general purpose
• easy to add new features
• Open standards
• promotes interoperability
• Highly cost-effective network hardware
• compared to traditional telephone switches

9
But...

• As of 5/2003, there were 1.2 billion cell phone
  subscribers in the world
• up 200 million in one year
• There are 4 billion IPv4 addresses, so in theory
  there's room
• but allocation issues make this impractical

10
What About NATs/DHCP?

• NATs preclude servers, or at least make them very
  cumbersome
• and we want to promote end-to-end architectures,
  not just wireless websurfing and mail reading
• DHCP only helps when most nodes are off
• yet everybody leaves their cell phones on
• Abolishing NATs and DHCP is what IPv6 is all
  about!
• and restore the original end-to-end architecture

11
My Vision

• All cellular networks become generic IPv6 ISPs
• Voice traffic transitions to IPv6 using open
  standards (SIP, freely usable codecs)
• Transparent interoperability and mobility between
  different wireless and wired technologies
• End-to-end encryption becomes routine
• Carriers compete on price, coverage, reliability,
  latency, etc
• may the best physical technology win

12
Cellular VoIP Obstacles

• Political
• Technical
• Security

13
Political Obstacles

• Telephone companies have always hated the
  Internet model
• they don't like being dumb-pipe providers
• they invented the buzzphrase "integrated
  services"
• they like captive customers (why do they fight
  number portability tooth and nail?)
• they still haven't grokked the end-to-end model
  (or maybe they have...)
• Their vendors like selling them expensive hardware

14
Technical Obstacles to VoIP

• Link Efficiency
• easily solved with header compression
• Standards/Interoperability
• VoIP world is still a mess (H323 vs SIP many
  proprietary voice codecs)
• Security
• denial of service spam worms, trojans
• potentially the biggest problem

15
The DoS Problem

• In a true IP architecture, every phone will be an
  Internet server with a global (IPv6) address
• any host anywhere can send it packets
• Wireless is inherently slower than wired
• Denial-of-service attacks would be too easy
• already pandemic in the wired Internet
• excess capacity keeps them from being more
  destructive than they already are

16
Blocking DoS Attacks

• Filters in the phone won't work
• the damage is to the wireless link, not the phone
• I.e., filters have to be in the network
• but under the control of the end-user
• open standards are required
• This problem isn't unique to wireless hosts
• they are simply the most vulnerable
• we need a general solution for all hosts if IPv6
  is to restore the end-to-end model

17
Blocking Spam

• Special class of denial-of-service attack
• attacked resource is user's eyes, not his link
• already a serious problem with SMS in some areas
• Best solution so far content-based filtering,
  e.g., Bayesian analysis, performed upstream under
  user control

18
Summary

• IP is the answer (IHMO) to the cellular telephony
  interoperability problem
• Only IPv6 will scale while keeping the end-to-end
  architecture that made the Internet great
• cellular phones could be IPv6's 'killer ap'
• Security is the biggest challenge, and will
  require a lot of careful, up-front attention

19
Thank You