intERLab

1
DUMBO an Emergency Network in Action Kanchana
Kanchanasut Computer Science School of
Engineering and Technology Asian Institute of
Technology Hajemi Tazaki WIDE Project,
Japan Witt Hmone Interlab, Asian Institute of
Technology
2
Background

• During times of disasters, such as an earthquake
  or tsunami, communications infrastructure has
  most likely been damaged or destroyed.
• Rescue and recovery efforts are usually hampered
  by communications failure.
• Ad-hoc communication facilities must be quickly
  put in place to support the command, control and
  coordination in the rescue and recovery
  operations.

3
Plan

• Introduction
• Emergency Network
• About the project
• Design considerations
• Physical components
• Networks
• Mesh vs Centralized
• OLSR vs AODV
• Experimental results
• Measurements and Analysis
• Conclusions

4
Plan

• Introduction
• Emergency Network
• About the project
• Design considerations
• Physical components
• Networks
• Mesh vs Centralized
• OLSR vs AODV
• Experimental results
• Measurements and Analysis
• Conclusions

5
Post-Disaster Crisis Management

• Collaboration and Communications are the key
• Can we always rely on existing communication
  infrastructure?

Source http//www.molokaihealthguide.com/images/
6
Disaster and Infrastructure
Mobile Switching Center (MSC)
Public Switched Telephone Network
BS
Local Branch
Local Branch
BS
BS
7
Telephony Networks Congestion or Damage
Mobile Switching Center (MSC)
Public Switched Telephone Network
BS
Local Branch
Local Branch
BS
BS
8
Hurricane Katrina

• After surviving Hurricane Katrina's initial
  blow, the radio communications system for the New
  Orleans police and fire departments dissolved as
  its radio towers lost their backup power
  generators in the ensuing flood.
• Some of the equipment could have been brought
  back up quickly, except that technicians were
  blocked from entering the submerged city for
  three days by state troopers who were themselves
  struggling with an overwhelmed radio system from
  a different manufacturer.
• With regular phone and cellular service knocked
  out in Katrina's wake -- the New Orleans mayor's
  office had to cobble together an Internet phone
  link with the outside world -- first responders
  were simply unable to share essential
  information. -- LiveScience.com

9
Communication in Emergency Scenario

• Traditionally walkie-talkies (Push-to-Talk) or
  two-way terrestrial radios have been the choice
  for rescue scenarios
• Push-to-Talk (PTT) service broadcasts voice
  information in short bursts to all receivers in
  the channel.
• PTT provides robust analog voice communication
  only for short to medium ranges, up to several
  kilometers.
• No multimedia content and not a networking
  solution.

10
Why multimedia?

• Post-Disaster Communications
• Text, e-mail and instant messaging
• Map, location information and GIS
• Graphics, images and Video

11
Why a network solution?

• A network solution is necessary to provide
  services like
• e-mail, instant messages, video/voice,
  information sharing and geographical location
  information thus enable delivery of
  mission-critical multimedia data between rescue
  team members locally and their headquarter over
  extremely long distances.
• Data network is interoperable with the global
  internet.

12
About DUMBO Project

• Interlab, AIT, Thailand
• Hipercom Project, INRIA, France
• WIDE Project, Japan
• With partners
• I2R (Singapore), Live!E Project (Japan)
• DUMBO I demonstration Dec 2006
• DUMBO II demonstration Oct 2008

13
Plan

• Introduction
• Emergency Network
• About the project
• Design considerations
• Physical components
• Networks
• Mesh vs Centralized
• OLSR vs AODV
• Experimental results
• Measurements and Analysis
• Conclusions

14
Design Considerations

• Use of day-to-day equipments laptops, PC and
  PDAs
• Minimum technical training required
  self-configuring/healing
• Able to meet the diversity of requirements in the
  emergency situation, thus wireless solution
• Long range communication by satellite links or
  WiMax
• For the short range requirement, commercial
  solutions with widely available IEEE 802.11
• Multimedia Applications
• Rich conversations, sensor integration, face
  recognition

15
DUMBO06Wireless Mesh Network and satellite
Internet Satellite
Field Satellite Access
IP Star Uplink
MANET OLSR
IP Star Gateway
Field Satellite Access
Terrestrial Internet
MANET OLSR
Simulated Disaster Area 2
sensor
Simulated Head Command Center (AIT)
Simulated Disaster Area 1
16
Mesh vs. Centralized

• WiMax vs mesh WiFi?

17
Two conflicting schools
Access Point
wireless centralized scheme
wireless meshed scheme

• Ref Philippe Jacquet, HIPERCOM Project,INRIA

18
Facts and figures
mesh with node density 10,000
capacity ratio per ha
mesh with node density 1,000
multi-centralized scheme
mesh with node density 100

• Ref Philippe Jacquet,
• HIPERCOM. INRIA

access point density
19

• Centralized schemes and mesh schemes have same
  interference problems
• Mesh schemes scale better
• offer better flexibility
• adapt to and benefit from node density

• Ref Philippe Jacquet, HIPERCOM Project,INRIA

20
OLSR vs AODV

• Choice of Mobile Ad Hoc Network Protocol
• Optimized Link State Routing
• or
• Ad Hoc On-Demand Distance Vector

21
OLSR

• Though consume resources in keeping up-to-date
  routing information, it is more efficient of
  sporadic network traffic with less overhead
• Cross-layered applications sharing Hello and
  Topology Information control messages

22
OLSR

• IETF RFC 3626
• Table-driven proactive protocol
• OLSR reduces control traffic by using Multipoint
  Relays (MPRs) instead of pure flooding the
  network
• Network topology maintained periodically
• Implementations (Open source)
• INRIA (http//hipercom.inria.fr/olsr/ )
• Unik ( http//www.olsr.org )
• NRL (http//cs.itd.nrl.navy.mil/work/olsr/index.ph
  p )

23
Plan

• Introduction
• Emergency Network
• About the project
• Design considerations
• Physical components
• Networks
• Mesh vs Centralized
• OLSR vs AODV
• Experimental results
• Measurements and Analysis
• Conclusions

24
Multimedia Communications for Disaster Emergency
Responses

• Interactive Video, Voice, and Instant Messaging
• very important to situational awareness
• Peer-to-Peer Paradigm ( no centralized server )

25
Experimental Testbed
26
Climate Sensors
Courtesy of Live E!
27
Face Recognition System
Face Recognition to search and identify people
28
Test Scenarios

• The testbed experimentation is divided into two
  scenarios
• Intra-team communications between nodes within
  the same MANET (referred to as Testbed Scenario A
  or just Scenario A)
• and Inter-team communications between nodes in
  different MANETs (referred to as Testbed Scenario
  B or just Scenario B).

29
Throughput from 4 streams of Video to HQ
30
1. VOIP measurement and analysisby Ge Yu Julia,
I2R

• PING results and E-model
• We can clearly see that only the G.711 codec
  produces a toll quality in Scenario A, and when
  we add Packet Loss Concealment (PLC), the quality
  improves by 20.
• The results in Scenario B (satellite link
  included) cannot meet the requirement of VoIP.
• However, for Scenario B, Push-to-talk (PTT) is a
  possible solution, because the PTT has relaxed
  requirements compared to typical interactive
  VoIP.

31
2. MPR selection

• MPR selection problem across a high-delay bridged
  VPN.
• Node A may see M and N as its 1-hop neighbor if
  we bridge M and N through a high-delay link.

32
3. Jitter

• Audio -- Not very different on path B and C
• Video -- Significant different

33
Plan

• Introduction
• Emergency Network
• About the project
• Design considerations
• Physical components
• Networks
• Mesh vs Centralized
• OLSR vs AODV
• Experimental results
• Measurements and Analysis
• Conclusions

34
What have we learnt?

• OLSR with long delay link
• Need MPR selection modification
• Observations
• WiFi equipements have a proclivity to connect
  with equipments of the same type
• Packet loss rates varies significantly with
  devices in use
• Next
• Connection to the Internet
• MANET and NEMO
• Cross-layered P2P application

35
Thank You
Postal AddressP.O. Box 4, Klong
LuangPathumthani 12120,ThailandDirect
line(66-2)524-6611Fax[phone]Email
info at interlab.ait.ac.thhttp//www.interlab.ait
.ac.th