Wiki
Clone wikignrs / Minutes 2012-09-20
Status Report
- Delay Module is complete, early testing (virtual machine) shows acceptable performance. Further testing will be performed on either Orbit nodes or Trillian machines.
- Found some papers discussing mobility models for humans on cell phones. Predictive models show movement between locations as well as time spent at a single location.
Discussion
Human Mobility Modeling
- What models will we use for the client simulation.
- What models will be used for predicting TTL?
- Need to evaluate errors in both models.
- Is a flat address space or LNRS-style hierarchy appropriate and how does this impact our performance?
New Tasks
- Pin the entire GNRS server process to a single processor core for evaluation. Give-up on multi-thread/multi-core performance metrics.
- Work on simple cache implementation.
- GNRS "shortest-path" selection based on outbound delay table.
Updated