3. Opportunistic
Persistent Data
Storage
Weerasooriya
W.A.A.C.P.
Index:11002192
Supervised by:
Dr. Ranasinghe
D.N.
Introduction
Our Aim
Research
Questions
Our Study
Evaluation
Opportunistic Networks
Theorem (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.
4. Opportunistic
Persistent Data
Storage
Weerasooriya
W.A.A.C.P.
Index:11002192
Supervised by:
Dr. Ranasinghe
D.N.
Introduction
Our Aim
Research
Questions
Our Study
Evaluation
Opportunistic Networks Cont.
The base of the modern opportunistic network is the
Interplanetary Internet research project led by Vint
Cerf in 1998 at NASA.
This idea has grown rapidly and today a person with
a mobile device can be considered as a node in
opportunistic network which exhibits social network
properties.
8. Opportunistic
Persistent Data
Storage
Weerasooriya
W.A.A.C.P.
Index:11002192
Supervised by:
Dr. Ranasinghe
D.N.
Introduction
Our Aim
Research
Questions
Our Study
Evaluation
Social Network Properties
Opportunistic networks exhibit social network
properties.
There are some nodes which have a large number of
connections while others have few.
Popularity of those nodes is very high.
Finding such nodes in a network is beneficial to
deliver data as well as to store data.
10. Opportunistic
Persistent Data
Storage
Weerasooriya
W.A.A.C.P.
Index:11002192
Supervised by:
Dr. Ranasinghe
D.N.
Introduction
Our Aim
Research
Questions
Our Study
Evaluation
Opportunistic Persistent Data Storage
Many studies on opportunistic routing,content
distribution have been done.
But it is very hard to find studies on persistently
storing data in opportunistic networks.
Our focus is to implement a persistent data storage
on opportunistic network efficiently with the help of
social network properties.
12. Opportunistic
Persistent Data
Storage
Weerasooriya
W.A.A.C.P.
Index:11002192
Supervised by:
Dr. Ranasinghe
D.N.
Introduction
Our Aim
Research
Questions
Our Study
Evaluation
Research Questions
How can we implement a Storage
protocol(read/write/take) efficiently using social
network properties.
While doing that replicas should be kept and
handled energy efficiently with less overhead in the
opportunistic environment.
Extending the tasks in to streaming data.
14. Opportunistic
Persistent Data
Storage
Weerasooriya
W.A.A.C.P.
Index:11002192
Supervised by:
Dr. Ranasinghe
D.N.
Introduction
Our Aim
Research
Questions
Our Study
Evaluation
Our Study
After going through enough literature we have some
basic idea on how to do that.
We could find few implementations on reliable
storages in Mobile Ad-Hoc.
We have to study and find efficient way of using
social network properties to increase the reliability.
Figure : Storage model
15. Opportunistic
Persistent Data
Storage
Weerasooriya
W.A.A.C.P.
Index:11002192
Supervised by:
Dr. Ranasinghe
D.N.
Introduction
Our Aim
Research
Questions
Our Study
Evaluation
Our Study Continued
In the network we have to select a set of nodes as
servers and other nodes become clients.
We can use social network properties when
determining the server nodes.
Clients can interact with one of the server
nodes.(Agent)
There should be some storage to store messages in
each server
We should not write to all the sever nodes.Selected
servers may be used(write quorum)
Same thing for the read-(read quorum)
17. Opportunistic
Persistent Data
Storage
Weerasooriya
W.A.A.C.P.
Index:11002192
Supervised by:
Dr. Ranasinghe
D.N.
Introduction
Our Aim
Research
Questions
Our Study
Evaluation
Our Study Continued
Replicas are there to increase the availability of the
data.
Due to high mobility in opportunistic networks,
some hot data may be kept more than one places.
When using replicas the other important factor is
the correctness of the replicated data i.e. ”Is this
the latest data? or it’s an old replication?”
There should be a mechanism to handle this.
We can have a some unique identifiers for messages
mid (source id, object id,version id) and each node
can keep the mid list.
Older id’s can be replaced with new ones.
19. Opportunistic
Persistent Data
Storage
Weerasooriya
W.A.A.C.P.
Index:11002192
Supervised by:
Dr. Ranasinghe
D.N.
Introduction
Our Aim
Research
Questions
Our Study
Evaluation
Our Study-Summery
We should use social network properties that can be
seen in opportunistic networks to improve the
performance of storage protocol as well as the
replication mechanism.
It may be beneficial to know the most popular node
set in the opportunistic network.They may be
potential candidates for servers.
There are few matrices to find popular nodes in an
opportunistic network.Ex:Betweenness Centrality.
21. Opportunistic
Persistent Data
Storage
Weerasooriya
W.A.A.C.P.
Index:11002192
Supervised by:
Dr. Ranasinghe
D.N.
Introduction
Our Aim
Research
Questions
Our Study
Evaluation
Evaluation
Handling real world traces for opportunistic network
evaluations is very costly.
We have to cover large geographic are, have to find
some people (atleast 50).
Finding devices, place are other problems.
Therefore we opted to synthetic traces.
Finding a proper simulator for human
behaviour/opportunistic networks was very difficult.
22. Opportunistic
Persistent Data
Storage
Weerasooriya
W.A.A.C.P.
Index:11002192
Supervised by:
Dr. Ranasinghe
D.N.
Introduction
Our Aim
Research
Questions
Our Study
Evaluation
ONE simulator
This has been designed specially for Opportunistic
Networks.Developed using JAVA1
The mobility model that we use here is very close to
human behaviour in opportunistic networks(Working
Day Movement Model).
We can add our own mobility model or modify any
mobility model as we want.
There are six routing protocols.What we focus to
use is MaxPop.Like mobility models we can use our
own routing algorithm.
Capable of providing post processing reports.It will
be beneficial when analysing results.
1
http://www.netlab.tkk.fi/tutkimus/dtn/theone/
24. Opportunistic
Persistent Data
Storage
Weerasooriya
W.A.A.C.P.
Index:11002192
Supervised by:
Dr. Ranasinghe
D.N.
Introduction
Our Aim
Research
Questions
Our Study
Evaluation
Assumptions
We do not consider security as an issue when
sharing contents over the mobile devices.
No specific hardware or software restrictions for the
communication other than the things that are
considered in the project.
No peer failures occur during the communication.
Number of members in the opportunistic network
are fixed.
26. Opportunistic
Persistent Data
Storage
Weerasooriya
W.A.A.C.P.
Index:11002192
Supervised by:
Dr. Ranasinghe
D.N.
Introduction
Our Aim
Research
Questions
Our Study
Evaluation
References
Luo,J.,Hubaux,J.-P., Eugster, P. T. (2003). PAN:
Providing Reliable Storage in Mobile Ad Hoc
Networks with Probabilistic Quorum Systems. the
4th ACM MobiHoc, 1–12.
Pongthawornkamol, T., Nahrstedt, K., Wang, G.
(2007). The Analysis of Publish/Subscribe Systems
over Mobile Wireless Ad Hoc Networks. Mobile and
Ubiquitous Systems: Networking Services, 2007.
MobiQuitous 2007. Fourth Annual International
Conference on, 1–8.
Nanda, S., Kotz, D. (2008). Q. Bridging Centrality.
Proceedings - International Conference on Computer
Communications and Networks, ICCCN, 62–67.
27. Opportunistic
Persistent Data
Storage
Weerasooriya
W.A.A.C.P.
Index:11002192
Supervised by:
Dr. Ranasinghe
D.N.
Introduction
Our Aim
Research
Questions
Our Study
Evaluation
References
Ker¨anen, A. (2009). R. ONE simulator. Proceedings
of the Second International ICST Conference on
Simulation Tools and Techniques, 55.
Padmanabhan, P., Gruenwald, L., Vallur, A.,
Atiquzzaman, M. (2008). A survey of data
replication techniques for mobile ad hoc network
databases. The VLDB Journal.