Download - Assessment Model for Opportunistic Routing
Waldir Moreira, Paulo Mendes, and Susana Sargento [email protected]
Oct 24th, 2011IEEE Latincom 2011, Belém-PA/Brasil
Assessment Model for Opportunistic Routing
2
Agenda
• Introduction
• Carried Analysis
• Our Work
• Evaluation and Results
• Conclusions and Future Work
3
Introduction
• Powerful devices
• Spontaneous networks
• Opportunistic contacts
- Intermittent connectivity
• Many routing solutions
- Epidemic, encounter history, social aspects ...
• Different classifications
- Identify different families
- Application requirements to algorithm mapping
• Different performance metrics and experimental setups
4
Motivation
• Clear classification of solutions
• Fair way to assess routing performance
- Importance of a homogeneous evaluation
- Parameter setups and performance metrics can favour some proposals
5
Goal
6
Close Look atDifferent Proposals
7
Performance Metric Identification
8
Existing Classifications
• Identify common aspects of solutions
- Routing strategy and metrics
9
Proposed Taxonomy
• Performance metrics
- Delivery probability, cost, and delay
• Experimental setup
- Network density (area, # of nodes, ...)
- Traffic (sources/destinations, load, ...)
10
Universal Evaluation Framework
• Heterogenous mobility
- Human (Working Day Movement)
- Veicular (Bus Movement)
- Random (Shortest Path Map-Based Mov.)
• Nodes belonging to different home/office/entertainment location
• Epidemic, PROPHET, and BubbleRap
11
Scenario
• Epidemic vs. PROPHET
- Better performance
- ~2.1x for Epidemic
- ~2.6x for PROPHET
- From 22% to ~46%
12
Results
• Epidemic vs. PROPHET
- More forwardings
- Over 44x for Epidemic
- Over 34x for PROPHET
- From 12% to 47%
13
Results
• Epidemic vs. PROPHET
- Greater delay
- ~5.5x for Epidemic
- ~9.5x for PROPHET
- From ~27% to ~37%
14
Results
• PROPHET vs. BubbleRap
- Better performance
- 55% for PROPHET
- 33% for BubbleRap
- Over 17 perc. points
15
Results
• PROPHET vs. BubbleRap
- Cost from ~40% to ~70%
16
Results
• For fair assessment
- Imperative to consider similar conditions
- Same performance metrics
• Both taxonomy and UEF should be updated to comply with future new trends
17
Conclusions andFuture Work
• To FCT for financial support via PhD grant (SFRH/BD/62761/2009) and UCR project (PTDC/EEA-TEL/103637/2008)
18
Acknowledgements
Waldir Moreira, Paulo Mendes, and Susana Sargento [email protected]
Oct 24th, 2011IEEE Latincom 2011, Belém-PA/Brasil
Assessment Model for Opportunistic Routing