Adaptive Mobilityassisted Data Dissemination in Mobile DisasterRecovery Environments

1
Adaptive Mobility-assisted Data Dissemination in
Mobile Disaster/Recovery Environments
Long Vu and Klara Nahrstedt University of
Illinois at Urbana-Champaign
longvu2,klara_at_cs.uiuc.edu
2
Motivation Disaster and Recovery
How to disseminate urgent messages to all
people in this area?
3
Motivation Heterogeneous Population Distribution
How to adaptively disseminate messages
according to network density?
4
Problem Statement

• Objectives
• Maximize coverage of message dissemination in the
  network
• Adapt data dissemination according to network
  node density
• Constraints
• Limited bandwidth network
• Mobile nodes
• Partitioned/disconnected network

5
System Models

• Network Model
• Mobile Ad Hoc Network mobile wireless devices
  carried by rescue team members and people
• Wireless nodes use the same channel
• Data Model
• Message alert, notification, etc.
• Priority high/low (urgent alert is more
  important than normal message)
• Deadline after its deadline, message becomes
  irrelevant (e.g. deadline of a message mes is
  1hour)

6
Adaptive Mobility-assisted Data Dissemination

• Importance Score to rank messages
• The top-k messages are broadcasted
• Adaptively tune broadcast period according to
  node density

7
Adaptive Mobility-assisted Data Dissemination
8
Message Manager

• Message Collection
• Operators add, update, delete, rank
• Ranks messages according to their Importance
  Scores
• Importance Score Calculator
• Periodically calculates Importance Scores of
  messages

9
Importance Score

• Recall our goal is message coverage

Imp(mes) Pri(mes) Time2Dline(mes) Dup(mes)
- Pri(mes) mes.priority / MAXPRIORIY -
Time2Dline(mes) mes.deadline (curTime
arrTime) / MAXDEADLINE - Dup(mes)
mes.duplication / MAXDUPLICATION

• Intuition
• - We prefer to disseminate higher priority,
  tighter deadline, and less duplicated message

10
Tuning Broadcast Period (1)
Node moves into areas with different node
densities
Neighbor list changes differently
Surrounding nodes move with different speeds
11
Tuning Broadcast Period (2)
A number of new neighbors in the (i-1)th
broadcast period Ti-1 B number of all alive
neighbors up to the (i-1)th broadcast period Ti-1
with
Initially
12
Evaluation Settings

• Mobility model Random Way Point
• Number of nodes 60 (max node speed 0.05m/s)
• Area of interest 50x50 (m2)
• Deadline of messages 300..500 (seconds)
• Transmission range 6,7,8 (meters)
• Sending buffer (SB or node memory) 40
  (messages)
• k 25, 35, 50 of Sending Buffer
• 15 random nodes create 200 messages

13
Importance Score Differentiates Messages
A message mes meets deadline if mes covers at
least 90 of nodes by its deadline
14
Tunable Broadcast Period Improves Meet Deadline
Tunable Broadcast Period
Fixed Broadcast Period
15
Conclusion

• Adaptive Mobility-assisted data dissemination
  scheme obtains message coverage under network
  bandwidth limitation
• Importance Score differentiates messages
• Broadcasts the top-k messages
• Gives preference to more important messages
• Tunable Broadcast Period improves performance
• More messages meet deadline

16
Variation of alpha