Untangling the Wireless Web

1
Untangling the Wireless Web

• Andrew .T. Campbell
• Columbia University
• comet.columbia.edu/campbell

2
Which Way to Wireless Internet?
Celcos viewpoint

• 2/2.5/
• GPRS, EDGE
• UMTS,WCDMA
• Cdma2000, 3X, HDR, 1Xtreme
• 3GPP, 3GPP2
• MWIF
• Bluetooth
• WAP, imode
• etc...

• Wireless LANs, PANS, MANETs
• Mobile IP
• Fast and seamless handoff
• Wireless IP QOS
• Fast security/AAA
• IP Paging
• TCP/IP Applications

4G?
3
3G Blues

• Insufficient bandwidth and service
  differentiation
• Spectrum poverty
• Cost of build out is prohibitive
• IP translators are the wrong application and
  service model (e.g., WAP, imode, etc.)
• Wrong architectural model for IP data

4
3G is the Wrong Model for IP

• Wrong network architecture to interconnect to
  Internet
• Need to be able to simply connect node Bs to the
  Internet
• Built on the circuit model regardless of packet
  service proposals its circuits!
• Radio Access Network, Physical Layer, Radio
  Interface Protocol and Radio Resource Management
• Complex/expensive centralized infrastructure
  (e.g., MSCs, RNCs)
• Strictly based on hierarchical networks
• Cant deliver the QOS model for data and
  multimedia?

• Each link layer implements specialized mechanisms
  which are similar for GSM, cdma2000, W-CDMA
• registration, authentication, paging, handoff,
  channel access, power control, frame selection

5
Positions and Challenges for 4G

• Wireless Data is king
• Direction design for data and IP - not voice
  circuits
• Wide variety of wireless link technologies
  2G/3G, wireless LAN, MANETs, PANs, wireless
  sensor networks new radio 4G technologies will
  emerge
• Direction IP over every radio link
• Multitude of mobile devices sensors, watches,
  pagers, pocket PCs, etc.
• Direction IP on every mobile device

6
Positions and Challenges for 4G

• Access to all Internet Applications (not filtered
  or restricted by service provider through WAP or
  imode)
• Direction Open access to all IP apps
• Bandwidth hungry apps and service differentiation
• Direction need to provide more bandwidth than 3G
  and resolve wireless IP QOS along the lines of
  differentiated services
• Demands for new wireless services
• Direction Design for dynamic service creation
  and deployment and exploit software radio ideas
  towards programmable mobile networks

7
Building 4G Infrastructure
Mobile IP enable Internet
Wireless VPNS
Managed Network
IP micro-mobility access networks
mobile ad hoc, PAN and Sensor networks
mobile device
8
My Favorite 4G

• Radio technology
• IEEE 802.11a
• IP Micro-Mobility
• Fast and seamless handoff
• IP Paging
• Wireless IP QOS
• Fast security/AAA services
• IP Protocols, Services and Apps
• Service Creation Middleware

9
Cellular IP

• Protocol
• Fast and seamless handoff control
• IP paging
• Fast security
• Wireless IP QOS
• Techniques
• Distributes mobile-aware functions (e.g., costly
  MSC kit)
• No new packet formats, encapsulation or address
  space
• Hosts are identified by their home addresses and
  packets are routed without tunneling or address
  conversion
• Minimizes explicit signaling and uses snooping to
  manage per-host soft-state routing and paging
  cache
• Uplink packets are routed to the gateway on a
  hop-by-hop basis
• Downlink packets a routed on the reverse path

10
Cellular IP Handoff

• Per-host routing for better real-time control
• Fast and semi-soft handoff operational modes
• Fast handoff
• simple but some loss experienced
• Semi-soft handoff
• Buffers packets using a delay device at
  cross-over router

11
Cellular IP Paging

• Supports scalability through separation of
  location management between idle/active mobile
  hosts
• Saves power, signaling and bandwidth
• supports paging areas and configurable paging
  cache
• Implicit IP paging approach
• Configurable paging cache supports directed
  broadcast

12
Wireless Differentiated Services

• Is it possible to achieve service
  differentiation using exclusively distributed
  algorithms?
• IEEE INFOCOM 2001
• Modified 802.11 DCF MAC to support soft
  prioritization
• Passive channel quality monitoring in the mobile
  host
• Estimation of achievable service quality
• Distributed admission control
• Fast SLA management during handoffs

13
Passive Virtual MAC

• Virtual MAC emulates a real 802.11 MAC
• receives virtual packets from a virtual
  application
• chooses random defer time
• observes channel and decrements timer
• if timer expires, transmits
• estimates collisions by observing transmission
  from other hosts
• adjusts CWmin, retransmits, etc...

collision!
success!
Radio busy
packet
decr. timer
decr. timer
delay of virtual packet
14
Simple Service Differentiation

• Service differentiation is statistical and
  relative
• High priority packets experience lower delay
• Channel capacity is efficiently shared between
  classes

Throughput
Delay
utilization
utilization
15
Distributed Admission Control

• Virtual-MAC delay estimates are precise
• An example algorithm
• A V-MAC estimates delay before starting session
• If estimated delay exceeds a conservative value,
  reject
• More sophisticated algorithms are under
  investigation
• Admission control based on higher moments
• Load control of LP together with admission
  control for HP

operating region
16
Programmable Mobile Networks

• Service creation environment
• Help automate the design, deployment, management
  and refinement of new network services
• migration paths to new technologies and services
  that builds on the concept of software radios
  that can create wireless physical layers
  dynamically
• Extend programmability to the MAC, signaling
  plane, transport and application level processing
  in wireless access networks

CIP to Mobiware
Reflective handoff service
17
4G Challenges

• Much higher bandwidth than 3G
• Mixed 4G networks (but all IP)
• Lots of radios
• Mobile IP enabled Internet
• IP Micro-mobility access networks
• Mobile ad hoc networks
• Sensor and deeply embedded networks
• Mixed QOS requirements for 4G (but all IP)
• Web data (real-time/non real-time, microflows)
• Continuous media (mostly real-time)
• TCP data (bursty best effort)
• All these networks will be interconnected and
  have to deliver support to mixed media

18
Toward 4G
mobility management
location management
power control/ frame selector
AAA, Security
IP QOS
Wireless IP
3G radios
4G radios
802.11
802.15
19
Toward 4G
service creation middeware
mobility management
location management
power control
AAA, QOS, others
programmable control components
wireless IP
GPRS/GSM
cdma2000
W-CDMA
programmable MAC 4G software radio
others
802.11
802.15
Programmable Mobile Networks
20
Our Contribution
Mobiware
PARO and INSIGNIA
Cellular IP, CIMS, PMIP and Utility
fast seamless handoff
IP paging
IP QOS
QOS in MANET Power-ware Routing
programmable control components
wireless IP
GPRS/GSM
cdma2000
W-CDMA
programmable MAC 4G software radio
others
802.11
802.15
21
Wireless Programming Model
beaconing systems
beaconing systems
beaconing systems
measurement systems
beaconing systems
detection algorithms
beaconing systems
handoff adapters
physical/data link layers
Programmable MAC Software Radio
22
Virtual Wireless Networks on-demand

physical network
23
Programmable Wireless Internet
WISP
WISP
Wireless VPN Service Provider
Wireless Internet
WISP
WISP
24
More Information

• comet.columbia.edu/campbell
• Papers on CIP, wireless IP QOS and Programmable
  Mobile Nets
• comet.columbia.edu/micromobility
• Columbia IP Micro-mobility Suite (CIMS) ns-2
  exetension
• NS for Cellular IP, Hawaii and HMIP
• comet.columbia.edu/cellularip
• Cellular IP source code

25
Thanks for listening!