Reliable multicast scheme using a bottleneck location estimator

Abstract

Multi-media application services are sensitive to loss, delay and delay jitter. The router and link failure may induce the packet losses.There are many studies for reliable multicast service. The fundamental bases of reliable multicast protocols is loss recovery. Traditionally there are two different ways of recovering missing packets; Forward error correction (FEC)and Automatic repeat request(ARQ) or retransmission. The behind using ARQ is to retransmit a packet only if it is loss. The main advantage of this scheme is bandwidth conservation because on redundant packets are transmitted. However, the disadvantage of retransmission based recovery mechanism is the increase in end-to-end latency because at least a rout-trip time elapses between the time a receiver sends a retransmission request and actually receives the retransmitted packet from the sender. Forward error correction-based recovery takes a different approach. Redundant packets are sent together with regular data packets from the sender to the receiver. Depanding on how many redundant packets are sent and how many packets are lost at the receiver. It is possible to recover missing packets based on the received packets. Thus FES-based approach reduces the end -to-end latency at the cost of additional bandwidth However FEC alone is not enough to guarantee complete reliability in practice . In order to improve these problems, we can consider estimation mechanism before actual fault happens in nodes or links. Anoop proposed fault isolation mechanism which detects fault routes and links by using Mtrace(Multicast Traceroute). This technique uses path probe history, and shared learning of probe results to scale well and infer the location of route changes, router failure, or loss link. However the fault isolation need additional bandwidth to periodically trace multicast path from each group member to source. Also Sylvia showed multicast tree based inference model is applicable to estimate the bottleneck no...