TY - GEN
T1 - A framed packet switch without control loop
AU - Yu, Le
AU - Nong, Ge
AU - Hamdi, Mounir
PY - 2011
Y1 - 2011
N2 - In this paper, we propose a 3-stage framed packet switch using an internal speedup of 2 to avoid any control loop between any two stages of the switch. The switch segments the arriving variable-length packets at each input port into fixed-size cells and assembles the cells into frames. Then the frames are switched across the shared buffers to their destined output ports, and the cells are reassembled into packets before being transmitted to the next hop. We have designed a broad class of work-conserving scheduling algorithms for the proposed switch, and they are analyzed to be stable, i.e. achieving 100% throughput, under any admissible traffic. To gain more insights into the switch practical performance, an extensive performance evaluation study is conducted using computer simulations. Our results demonstrate that the worst-case performance can be bounded. In addition, we are able to achieve a high throughput-delay performance comparable to that of the padded frame switch which uses a much more complicated scheduling algorithm [1].
AB - In this paper, we propose a 3-stage framed packet switch using an internal speedup of 2 to avoid any control loop between any two stages of the switch. The switch segments the arriving variable-length packets at each input port into fixed-size cells and assembles the cells into frames. Then the frames are switched across the shared buffers to their destined output ports, and the cells are reassembled into packets before being transmitted to the next hop. We have designed a broad class of work-conserving scheduling algorithms for the proposed switch, and they are analyzed to be stable, i.e. achieving 100% throughput, under any admissible traffic. To gain more insights into the switch practical performance, an extensive performance evaluation study is conducted using computer simulations. Our results demonstrate that the worst-case performance can be bounded. In addition, we are able to achieve a high throughput-delay performance comparable to that of the padded frame switch which uses a much more complicated scheduling algorithm [1].
KW - 100% throughput
KW - 3-stage switching
KW - internal speedup
KW - load-balanced
UR - https://www.scopus.com/pages/publications/80052994478
U2 - 10.1109/ICCCN.2011.6006043
DO - 10.1109/ICCCN.2011.6006043
M3 - Conference contribution
AN - SCOPUS:80052994478
SN - 9781457706387
T3 - Proceedings - International Conference on Computer Communications and Networks, ICCCN
BT - 2011 20th International Conference on Computer Communications and Networks, ICCCN 2011 - Proceedings
T2 - 2011 20th International Conference on Computer Communications and Networks, ICCCN 2011
Y2 - 31 July 2011 through 4 August 2011
ER -