Presents the first results of my PhD proposal, which resulted in the following papers:
MOREIRA, W., SOUZA, M., MENDES, P., SARGENTO, S.
Study on the Effect of Network Dynamics on Opportunistic Routing.
In: Proceedings of the 11th International Conference on Ad-Hoc Networks and Wireless (AdHoc Now 2012), 2012, Belgrade, Serbia.
MOREIRA, W., MENDES, P., SARGENTO, S.
Opportunistic Routing Based on Daily Routines.
In: Proceedings of the 6th IEEE WoWMoM Workshop on Autonomic and Opportunistic Communications (AOC 2012), 2012, San Francisco, USA.
This presentation was given in SITI Brainstorming meeting, on Feb 1st, 2012 @ SITI.
4. Picture today
• Users are eager for retrieving/providing
information
• Popularization of portable devices
4
5. Straightforward Definition
OppNets are highly dynamic, composed of
mobile and static nodes (i.e., devices) and
take advantages of opportunistic time-
varying contacts among users carrying them
to exchange information
5
6. General OppNets
Characteristics
• Occasional contacts
• Intermittent connectivity
• Highly mobile and fixed nodes
• Power-constrained devices
• Possible nonexistence of e2e paths
6
8. Different Environments
• Disruptive environments:
- Sparse scenarios where communication
is established through sporadic contacts
• Urban environments
- Dense scenarios with communication
suffering different interference levels
8
9. Disruptive Environments
Deep Space Communications
• Purpose: provide communication means
for manned/robotic exploration
• Main challenges: very long delays,
sparseness, shadow areas and spacecraft
lifetime
• Function: Information and commands are
exchanged between landers/rovers and
earth station through orbiters
9
11. Disruptive Environments
Networks for Developing World
• Purpose: provide asynchronous Internet
access despite the scarce/expensive
infrastructure
• Main challenges: long delays and
scarce/expensive infrastructure
• Function: data is sent/retrieved either
through USB stick carried by a motorbiker
or via dial-up connection
11
12. Disruptive Environments
Networks for Developing World
[3] S. Jain, K. Fall, R. Patra, Routing in a delay tolerant network, 2004
[4] News on Pigeon Carrier
12
13. Disruptive Environments
Zebranet
• Purpose: Study zebra movements
through collars carried by them
• Main challenges: energy constraints
• Function: collars opportunistically
exchange GPS location later then
obtained by scientists
13
15. Disruptive Environments
Tactical Military Networks
• Purpose: establish quick communication
means among military soldiers, vehicles,
and aircrafts
• Main challenges: high disruption and
partition
• Function: information is relayed among
military units
15
17. Urban Environments
Opportunistic Sensing
• Purpose: gather information from sensing
systems
• Main challenges: short contact times
• Function: sensor present in different
devices gather information which is then
collected mobile devices (i.e., custodian)
to be transfered to the sensing system
central
17
20. What is it about?
Considers any contact among nodes and
forwarding decisions are made using locally
collected knowledge about node behavior to
predict which nodes are likely to deliver a
content or bring it closer to the destination
20
21. 2000-2010 Analysis
[7] W. Moreira and P. Mendes, “Survey on opportunistic routing for delay
tolerant networks,” SITI, University Lusofona, February, 2011
21
23. Major Routing Families
[7] W. Moreira and P. Mendes, “Survey on opportunistic routing for delay
tolerant networks,” SITI, University Lusofona, February, 2011
23
24. Social Aspects:
The New Trend
• Since 2007
• Have shown great potential
• Use social relationship
• Much wiser decisions
24
25. Replication-based Approaches
Social Similarity
• Community Detection: creation of communities
considering people social relationships
- Bubble Rap
* Forwarding based on
community and local/
global centrality
[8] P. Hui, J. Crowcroft, E. Yoneki, BUBBLE Rap: Social-based Forwarding in
Delay Tolerant Networks, 2011
25
26. Replication-based Approaches
Social Similarity
• Shared Interests: nodes with the same interest
as destination are good forwarders
- SocialCast
* predicted node’s co-location (probability of
nodes being co-located with others)
* change in degree of connectivity (mobility and
changes in neighbor sets)
[9] P. Costa, C. Mascolo, M. Musolesi, G. P. Picco, Socially-aware routing for
publish-subscribe in delay-tolerant mobile ad hoc networks, 2008
26
27. Replication-based Approaches
Social Similarity
• Node Popularity: use of social information
to generate ranks to nodes based on their
position on a social graph
- PeopleRank
* Forwarding based on social ranking of
nodes
[10] A. Mtibaa, M. May, M. Ammar, C. Diot, Peoplerank: Combining social
and contact information for opportunistic forwarding, 2010
27
29. Motivation
• Community detection, shared interests, node popularity
• Communities are statically defined
• Do not consider the age of contacts when computing the
centrality
• Strong assumptions
• Full knowledge on social information is not enough
• Some social metrics (e.g., betweenness centrality) can lead to
node homogeneity
[11] T. Hossmann, T. Spyropoulos, F. Legendre, Know thy neighbor:
Towards optimal mapping of contacts to social graphs for dtn routing, 2010
29
30. Our Proposal
dLife
People's daily life routine and their social ties
to reach a clean representation of social
interactions
Time-Evolving Contact Duration (TECD)
Weights social interactions based on statistical
contact duration nodes have over time
TECD Importance (TECDi)
Estimates the importance of nodes
30
32. Promising results
Average Delivery Probability
Created Scenario
1.0
0.8
0.6
0.4
0.2
Average Delivery Probability
0.0 Traces
1 day 2 day 4 day 1 week 3 week
TTL 1.0
0.8
0.6
BubbleRap
dLife Comm 0.4
dLife
0.2
0.0
1 day 2 day 4 day 1 week 3 week
TTL
32
33. Promising results
Average Cost
Created Scenario
1600
1400
1200
1000
# of replicas
800
600
400
200 Average Cost
0 Traces
1 day 2 day 4 day 1 week 3 week
TTL 40
35
30
25
# of replicas 20
BubbleRap
dLife Comm 15
dLife 10
5
0
1 day 2 day 4 day 1 week 3 week
TTL
33
34. Promising results
Average Latency
Created Scenario
45000
40000
35000
30000
Seconds
25000
20000
15000
10000 Average Latency
5000 Traces
0
1 day 2 day 4 day 1 week 3 week 45000
TTL 40000
35000
30000
Seconds 25000
BubbleRap
20000
dLife Comm
dLife 15000
10000
5000
0
1 day 2 day 4 day 1 week 3 week
TTL
34
35. Conclusions and Future Work
Functions in separate had good overall performance
Their combination sure provided improvements
dLife is able to transcribe the dynamic behavior
found on users' interactions into clean social
representations
Plans
Improve it by introducing randomness and a stale-
data removal scheme
35
36. References
[1] News on Deep Space Networking -
http://www.engadget.com/2008/11/19/nasas-interplanetary-internet-tests-a-
success-vint-cerf-triump/
[2] Mars Reconnaissance Orbiter -
http://www.nasa.gov/mission_pages/MRO/news/mro-20060912.html
[3] S. Jain, K. Fall, R. Patra, Routing in a delay tolerant network, in: Proceedings of
the ACM SIGCOMM, Portland, USA, August, 2004.
[4] News on Pigeon Carrier - http://www.dailymail.co.uk/news/article-
1212333/Pigeon-post-faster-South-Africas-Telkom.html
[5] MITRE Corporation (US Marine Corps) (Presentation on C2 On-the-Move
Network, Digital Over-the-Horizon Relay) -
http://www.ietf.org/proceedings/65/slides/DTNRG-2.pdf
[6] CamMobSens - Cambridge University Pollution Monitoring Initiative -
http://www.escience.cam.ac.uk/mobiledata/
36
37. References
[7] W. Moreira and P. Mendes, “Survey on opportunistic routing for delay tolerant
networks,” Tech. Rep. SITI-TR-11-02, Research Unit in Informatics Systems and
Technologies (SITI), University Lusofona, February, 2011.
[8] P. Hui, J. Crowcroft, E. Yoneki, BUBBLE Rap: Social-based Forwarding in Delay
Tolerant Networks, Mobile Computing, IEEE Transactions on, 10 (11)(2011) 1576–
1589.
[9] P. Costa, C. Mascolo, M. Musolesi, G. P. Picco, Socially-aware routing for publish-
subscribe in delay-tolerant mobile ad hoc networks, Selected Areas in
Communications, IEEE Journal on, 26 (5) (2008) 748–760.
[10] A. Mtibaa, M. May, M. Ammar, C. Diot, Peoplerank: Combining social and
contact information for opportunistic forwarding, in: Proceedings of INFOCOM,
San Diego, USA, March, 2010.
[11] T. Hossmann, T. Spyropoulos, F. Legendre, Know thy neighbor: Towards optimal
mapping of contacts to social graphs for dtn routing, in: Proceedings of IEEE
INFOCOM, San Diego, USA, March, 2010.
37
38. Using Social Information to
Improve Opportunistic Networking
Waldir Moreira
waldir.junior@ulusofona.pt
Feb. 1st, 2012
SITI Brainstorm Meeting