Name
Abstract | ||
|---|---|---|
Conventional multi-stage switches perform well in theory but degrade in application. The conventional traffic control to a flow corresponds to the destination module, so the flows destined to the same output module have bandwidth competition although they should not. The evaluation of conventional switches is only based on traffic patterns with these two features: the required bandwidth of all output ports is identical and no output is oversubscribed. Under their traffic patterns, bandwidth competition never happens. However, there exist other traffic patterns that they don't concern, in which the existed bandwidth competition will lead to degradation. We define this issue as Flow Interference (FI). To address it, we propose a philosophy that traffic control to a flow should correspond to its own destination output based on the global information. Based on this, we propose our switch architecture to solve FI using round-robin at all stages. Analysis and simulation results show that our multi-stage switch architecture can solve FI and achieve perfect performance, with O(1) computational complexity for the scheduling algorithm. |
Year | DOI | Venue |
|---|---|---|
2012 | 10.1109/APCC.2012.6388281 | APCC |
Keywords | Field | DocType |
bandwidth allocation,interference suppression,scheduling,telecommunication switching,telecommunication traffic,bandwidth competition,computational complexity,conventional switches,flow interference,multistage switch architecture,round-robin,scheduling algorithm,traffic control,traffic pattern | Bandwidth allocation,Scheduling (computing),Computer science,Flow (psychology),Computer network,Real-time computing,Bandwidth (signal processing),Interference (wave propagation),Dynamic bandwidth allocation,Network traffic control,Computational complexity theory | Conference |
ISSN | ISBN | Citations |
2163-0771 | 978-1-4673-4727-3 | 0 |
PageRank | References | Authors |
0.34 | 10 | 4 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Yupeng Tian | 1 | 0 | 0.34 |
| Xiaoping Zhang | 2 | 27 | 6.83 |
| Dehu Li | 3 | 0 | 0.34 |
| Chen Zhou | 4 | 6 | 0.87 |