TY - GEN
T1 - Reliable sender-initiated multicast for improved QoS
AU - Gupta, Vijay
AU - Campbell, Roy
N1 - Publisher Copyright:
© 2000 IEEE.
PY - 2000
Y1 - 2000
N2 - Network support for reliable sender-initiated multicast can aid software distribution, server-pushing of web-pages am well as pushing of prerecorded audio and video. Currently, bandwidth availability and propagation delays between the sender and the different recipients differ by many orders of magnitudes. Hence, an important problem which needs to be addressed for facilitating reliable sender-initiated multicast is this problem of network heterogeneity. Prior work in reliable multicast either presents solutions to transmit at the rate of the bottleneck link of the entire multicast tree, or assumes knowledge of static bandwidth to each of the recipients. We propose an algorithm which does not make these assumptionsbut partitions the set of recipients on the basis of the available bandwidths at the time when the multicast is started; and does a separate transmission to each set of recipients sharing a common quality of service (QoS). To the best of our knowledge,this is a flrst such solution. To achieve these goals, we propose (i) an algorithm to divide the set of receivers into classes with similar QoS, and (ii) callbacks for error recovery in reliable multicast. Our algorithm makes use of L4 switching at the routers but assumes the state at the routers to be soft state. Using analysis, we show our algorithm to be scalable for certain restricted network characteristics.
AB - Network support for reliable sender-initiated multicast can aid software distribution, server-pushing of web-pages am well as pushing of prerecorded audio and video. Currently, bandwidth availability and propagation delays between the sender and the different recipients differ by many orders of magnitudes. Hence, an important problem which needs to be addressed for facilitating reliable sender-initiated multicast is this problem of network heterogeneity. Prior work in reliable multicast either presents solutions to transmit at the rate of the bottleneck link of the entire multicast tree, or assumes knowledge of static bandwidth to each of the recipients. We propose an algorithm which does not make these assumptionsbut partitions the set of recipients on the basis of the available bandwidths at the time when the multicast is started; and does a separate transmission to each set of recipients sharing a common quality of service (QoS). To the best of our knowledge,this is a flrst such solution. To achieve these goals, we propose (i) an algorithm to divide the set of receivers into classes with similar QoS, and (ii) callbacks for error recovery in reliable multicast. Our algorithm makes use of L4 switching at the routers but assumes the state at the routers to be soft state. Using analysis, we show our algorithm to be scalable for certain restricted network characteristics.
UR - http://www.scopus.com/inward/record.url?scp=84963539137&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84963539137&partnerID=8YFLogxK
U2 - 10.1109/ICCCN.2000.885543
DO - 10.1109/ICCCN.2000.885543
M3 - Conference contribution
AN - SCOPUS:84963539137
T3 - Proceedings - International Conference on Computer Communications and Networks, ICCCN
SP - 551
EP - 556
BT - Proceedings - 9th International Conference on Computer Communications and Networks, ICCCN 2000
A2 - Engbersen, Ton
A2 - Park, E. K.
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 9th International Conference on Computer Communications and Networks, ICCCN 2000
Y2 - 16 October 2000 through 18 October 2000
ER -