Distributed link scheduling algorithms that obey routes inner-flow. (Networking)
Conducted byFabio Rocha Jimenez Vieira
DescriptionThese experiments are at least 100 jobs. Each one runs a simulated a random wireless network where our algorithm tries to reach theoretical throughput upper bounds defined by Gupta's work (Capacity of wireless networks). They use low amounts of memory and the resulting files are also small.
- Nodes involved: 500
- Sites involved: >3
- Minimum walltime: 8h
- Batch mode: yes
- Use kadeploy: no
- CPU bound: yes
- Memory bound: no
- Storage bound: no
- Network bound: no
- Interlink bound: no
Tools usedA particular wireless network simulator wrote in C++ (see publications).
ResultsWe proved a overall gain of 500% over classic schedule algorithms (see publications). We introduced 2 new algorithms (SERA and MRA).
Shared by: Fabio Rocha Jimenez Vieira
Last update: 2012-09-25 13:09:16