Traps and Pitfalls of Using Contact Traces in Performance...

13
Traps and Pitfalls of Using Contact Traces in Performance Studies of Opportunistic Networks Nikodin Ristanovic, George Theordorakopoulos and Jean-Yves Le Boudec.

Transcript of Traps and Pitfalls of Using Contact Traces in Performance...

Page 1: Traps and Pitfalls of Using Contact Traces in Performance …ey204/teaching/ACS/R202_2011_2012/... · 2012. 3. 6. · Traps and Pitfalls of Using Contact Traces in Performance Studies

Traps and Pitfalls of Using Contact Traces in Performance Studies of

Opportunistic Networks

Nikodin Ristanovic, George Theordorakopoulos and Jean-Yves Le Boudec.

Page 2: Traps and Pitfalls of Using Contact Traces in Performance …ey204/teaching/ACS/R202_2011_2012/... · 2012. 3. 6. · Traps and Pitfalls of Using Contact Traces in Performance Studies

Introduction•Background Research

•The Experiment

•Design goals

•Architecture

•Contact trace results

•Meta-data results

•Conclusions

Page 3: Traps and Pitfalls of Using Contact Traces in Performance …ey204/teaching/ACS/R202_2011_2012/... · 2012. 3. 6. · Traps and Pitfalls of Using Contact Traces in Performance Studies

Background Research

•Validation of simulation model results

•Contact-based opportunistic communication

• Effect of backbones in opportunistic communication

Page 4: Traps and Pitfalls of Using Contact Traces in Performance …ey204/teaching/ACS/R202_2011_2012/... · 2012. 3. 6. · Traps and Pitfalls of Using Contact Traces in Performance Studies

The Experiment - design goals

•Collect application data to extract performance metrics

•Collect contact traces to be used in event simulations

•Compare results with contact-based simulation values

Page 5: Traps and Pitfalls of Using Contact Traces in Performance …ey204/teaching/ACS/R202_2011_2012/... · 2012. 3. 6. · Traps and Pitfalls of Using Contact Traces in Performance Studies

The Experiment - architecture

•50 Roaming users (RU)

• 10 Home user equivalents (HUE)

• 1 Proxy

•Opportunistic twitter application

•Haggle pub/sub

• 3 RU Caching strategies

•Context-based HUE Caching strategy

Page 6: Traps and Pitfalls of Using Contact Traces in Performance …ey204/teaching/ACS/R202_2011_2012/... · 2012. 3. 6. · Traps and Pitfalls of Using Contact Traces in Performance Studies

The Experiment - architecture

Page 7: Traps and Pitfalls of Using Contact Traces in Performance …ey204/teaching/ACS/R202_2011_2012/... · 2012. 3. 6. · Traps and Pitfalls of Using Contact Traces in Performance Studies

Contact Trace Results

Page 8: Traps and Pitfalls of Using Contact Traces in Performance …ey204/teaching/ACS/R202_2011_2012/... · 2012. 3. 6. · Traps and Pitfalls of Using Contact Traces in Performance Studies

Contact trace Results

Page 9: Traps and Pitfalls of Using Contact Traces in Performance …ey204/teaching/ACS/R202_2011_2012/... · 2012. 3. 6. · Traps and Pitfalls of Using Contact Traces in Performance Studies

Meta-data Results

Page 10: Traps and Pitfalls of Using Contact Traces in Performance …ey204/teaching/ACS/R202_2011_2012/... · 2012. 3. 6. · Traps and Pitfalls of Using Contact Traces in Performance Studies

Meta-data Results

Page 11: Traps and Pitfalls of Using Contact Traces in Performance …ey204/teaching/ACS/R202_2011_2012/... · 2012. 3. 6. · Traps and Pitfalls of Using Contact Traces in Performance Studies

Meta-data Results

Page 12: Traps and Pitfalls of Using Contact Traces in Performance …ey204/teaching/ACS/R202_2011_2012/... · 2012. 3. 6. · Traps and Pitfalls of Using Contact Traces in Performance Studies

Conclusions

•Technological limitations need to be simulated

• Limiting cache size can aid architecture development

• Statistical predictions can aid system performance

Page 13: Traps and Pitfalls of Using Contact Traces in Performance …ey204/teaching/ACS/R202_2011_2012/... · 2012. 3. 6. · Traps and Pitfalls of Using Contact Traces in Performance Studies

Thanks

•Any Questions?