Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.

20
Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström

Transcript of Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.

Page 1: Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.

Challenged Networking

An Experimental Study of New Protocols and Architectures

Erik Nordström

Page 2: Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.

Unchallenged Networking

• Wired• Stationary• Conversational• Structured• End-to-end

Page 3: Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.

The Challenges

• Mobile computers• No structures• Intermittent connectivity• No end-to-end paths• No pre-established

names and addresses

How to enable communication when requirements and assumptions change?

Page 4: Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.

The Solutions

• Make existing protocols work better

• Develop new protocols• Make existing and new

protocols interoperable• Develop evaluation

methodologies and testbeds

Page 5: Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.

Main Contributions

• Experimental testbeds and methodologies– Ad hoc Protocol Evaluation (APE)

• Protocol implementation and evaluation– AODV-UU, DSR-UU routing protocols

• New communication paradigms– Interest-based dissemination

• New network architectures– Haggle

Page 6: Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.

Challenges of Experimental Evaluation

• Repeatability– Mobility– Radio

• Management– Scaling– Configuration

• Cross-environment– Simulation, emulation, real-world

Page 7: Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.

Related Work in TestbedsOrbit [Raychaudhuri et. al 2005]

DieselNet [Burgess et. al 2006]

[Above photo (c) 2006 Daily Hampshire Gazette]

•Netbed/Emulab [White et. al 2002]•Whynet [Zhou et. al 2005]•Etc.

Roofnet [Aguayo, et. al 2003]

Page 8: Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.
Page 9: Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.

Experiment Choreography1. Go to position B in the middle of the corridor (5 secs)

2. Continue to position C at the end of the corridor (5 secs)3. Go back to starting point (10 secs)

Log SNR

Log traffic

Log parameters

Page 10: Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.

0 1 2 3

Experimental Setup

Relay node swap

Ping!

Routing protocol implementations:•AODV-UU•DSR-UU•OOLSR

Page 11: Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.
Page 12: Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.

Thesis page 204

Link cache poisoning

Mean over 10 experiments

Page 13: Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.

0

3

0

3

Simulation

Real world

Source route at receiver

Source route at sender

Flip-flop routing

DSR-UU

DSR-UU

Source routes

Page 14: Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.

Haggle – An Architecture for Challenged Networking

• Scenario:– People carry information– Ad hoc/opportunistic interactions– Heterogeneous connectivity

• Problems:– How to agree on names and addresses?– How to exchange information?– How to agree on which information to exchange?

Page 15: Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.

The Haggle Approach: Searching

Interests

Interests

Search for matching content

Search for matching content

4 3 21

12 3 4

Page 16: Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.

Interest-based Dissemination

message

interest

Match at least two interests!

Page 17: Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.

Three Interest-based Dissemination Strategies

Wait – only disseminate from source

Content – intermediate nodes may re-disseminate

(forward)

Flood – any node may re-disseminate

Page 18: Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.

What is the Correlation between Contacts and Interests?

List of interests

AffiliationNationality

Technical interests…

3 cm

Page 19: Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.

Trace-based Simulation

1000 messages uniformly spread

over the trace.Match affiliation or

nationality.

Page 20: Challenged Networking An Experimental Study of New Protocols and Architectures Erik Nordström.

Conclusions

• Protocols and architectures for challenged networks cannot be designed solely based on simulations and emulations

• Protocols need to be designed for intermittent connectivity, but they should also exploit full connectivity

• Protocols need to rely more on decision making based on (long term) measurements of the environment, or on a-priori knowledge rather than lookup services

• New naming and addressing schemes required. Schemes that support searching can be attractive for flexible content exchanges