Building a Cellular IP Testbed

1
Building a Cellular IP Testbed

• Javier Gomez and Andrew .T. Campbell
• Columbia University
• comet.columbia.edu/wireless
• support from Ericsson, Intel, Nortel and IBM

2
Wireless IP views
3
3G Wireless

• Existing schemes
• IMT-2000
• too many proposals
• Pros
• smooth mobility support
• Cons
• circuit model
• complex/expensive infrastructure (e.g. MSCs)
• strictly based on hierarchical networks

4
Mobile IP

• Existing schemes
• IPv4 mobility with/without route optimization
• IPv6 mobility
• Pros
• Simple and scalable mobility solution
• Cons
• handoff latency and packet loss
• signaling load and performance scalability
• QOS

5
Micro-Mobility Schemes

• New proposals for fast handoff in IETF
• hierarchical foreign agents (Nokia,96)
• Cellular IP (Columbia/Ericsson,98)
• Hierarchical IPv6 (INRIA,98)
• HAWAII (Lucent,99)
• THEMA (Lucent/Nokia,99)
• Other initiatives
• ICEBERG (UCB/Ericsson)

6
Cellular IP Project

• Project started at Columbia with Ericsson, 97
• Simple Vision
• combining the strengths of Cellular IP
  without inheriting their weaknesses.
• Cellular IP inherits cellular technology
  principles
• but implements these around the IP paradigm
• Observation
• 3G and Internet fundamentally different

7
Design Goals

• fast and seamless handoff
• per-mobile routing soft-state
• real-time location tracking - implicit paging
• support for active and idle users
• passive connectivity
• single scalable protocol
• simplicity
• no new packet formats, encapsulation or address
  space
• distribute mobile-aware functions (e.g., costly
  MSC kit)
• built a foundation for QOS support

8
Macro/micro mobility
9
Building Blocks

• base station
• wireless access point
• routes IP packets
• integrated cellular control found in MSC and BSC
• IP routing replaced by Cellular IP routing
• gateways
• mobile IP support macro-mobility
• mobiles hosts attached to the network use the IP
  address of the gateway as their Mobile IP COA
• inside the network hosts are identified by their
  home addresses and packets are routed without
  tunneling or address conversion
• mobile host

10
MobileIP/CellularIP
11
Protocol Overview

• Location management handoff are integrated with
  routing
• data packets transmitted by a node are used to
  establish location and routing soft-state
• no explicit signaling is required
• uplink packets are routed to the gateway on a
  hop-by-hop basis
• downlink packets a routed on the reverse path
• idle mobiles allow state to timeout
• in-band paging mechanism locates idle hosts

12
Network Model

• Main Algorithms
• Routing, handoff and paging

13
Uplink Packets Shortest Path
14
Uplink packets Create Location Information
15
Hard Handoff

• Redirected uplink packets create new downlink
  path
• Optimal reuse of previous path

16
Semi-Soft Handoff

• Delay device for synchronization

home agent
E
C
R
G
Internet w/ Mobile IP
R
D
A
R
foreign agent
F
B
host
17
OK, Some control Messaging

• Control packets are regular IP packets with no
  payload
• They update routing entries
• Discarded before reaching the internet

18
Location Management of Idle Hosts

• paging mechanism
• in-band signaling for paging using live data
• target and broadcast technique

home agent
19
Security

• mobile and network have shared secret key
• control messages are authenticated
• Cache mappings cannot be created or modified by
  data packets
• control packets are ICMP
• control packets must contain timestamp and
  authentication information

20
Implementation Model
21
Cellular IP Testbed I
22
Cellular IP Testbed II
23
Hardware

• Cellular IP nodes/mobiles hosts Pentium 300 MHz
• wired links
• ethernet 10/100 Mbps
• wireless links
• Wavelan 2 Mbps
• Lucent 802.11, 2-11 Mbps
• Aironet 802.11, 2-11 Mbps

24
Downlink Packet Loss _at_Handoff
25
TCP Throughput _at_Handoff
26
Scalability

• Cellular IP uses per-host routes in order to
  support high performance handoff
• does N impacts the performance?

Cellular IP node
100 Mbps
100 Mbps
Test mobile
ttcp
N
27
cellular IP nodes throughput
28
Scalability

• The main scalability bottleneck is the overhead
  associated with life tracking of mobiles
• Cellular IP achieves scalability by
• separation of location management between
  idle/active mobile hosts

29
(Wireless LANs today/tomorrow?)

• Today
• target coverage and high throughput as primary
  design goals
• there is no provision for frequency reuse
• power control is difficult
• Tomorrow?
• data access still random (CDMA/CSMA)
• there is some sort of control channel...
• Improved frequency reuse/Handoff support
• slotted paging power savings

30
(No Transcript)
31
Summary

• 3G is being redefined toward IP centric solutions
• Number of new proposals on micro-mobility
• New 3GIP working group
• Cellular IP
• is capable of combining the strengths of
  Cellular and IP approaches without inheriting
  their weaknesses
• source code available November 99
• comet.Columbia.edu/cellularip
• DEMO Cellular IP at IEEE MOMUC99

32
Cellular IP Team

• Andrew T. Campbell
• Javier Gomez
• Sanghyo Kim
• Bill Paul
• Andras G. Valko (Ericsson Research)
• Zoltan Turanyi (Ericsson Research)
• Chieh-Yih Wan

33
Publications

• Papers
• "Design, Implementation and Evaluation of
  Cellular IP", IEEE Personal Communications,
  August 200.0
• Internet Draft
• Cellular IP, Internet Draft, draft-ietf-mobileip-c
  ellularip-00.txt, IETF Mobile IP Working Group
  Document, December 1999.
• Papers, IDs and source code
• comet.columbia.edu/cellularip/publications.htm