Design and evaluation of a QoS-adaptive system for reliable multicasting

Learning from the past for resolving dilemmas of asynchrony

“…Any p j that is transmitting m halts the activity if a higher ranked p j' is observed to be doing the same. Our simulations in [8] indicate that the suppression mechanism is effective in reducing message cost from O(n 2 ρ) to O(nρ).…”

“…The protocol is an extension of our earlier work [8] developed for reliable multicasting. For space reasons, we outline the order protocol and work out an analytical estimation of D for a given ∆ assuming single-packet messages (which are like the probe messages).…”

A QoS-Negotiable Middleware System for Reliably Multicasting Messages of Arbitrary Size

“…The first is to purchase the CS services from an Internet Service Provider (ISP) such as AT&T. Our earlier work [1] assumed an ISP maintained CS with guaranteed QoS metrics. This approach however restricts the service construction/composition to be ISP-centric: while each ISP offers attractive QoS guarantees within its own domain, ISPs do not cooperate between themselves to offer such guarantees across their domains; consequently, the CS of 'networked system' is merely the domain of a single ISP.…”

“…Moreover, a multicast message can be of arbitrary (but known) size, which needs to be fragmented into several packets prior to its transmission. We accomplish this task by extending an earlier work in [1] which regards each message as a single packet. Extension will be done in two distinct ways by considering the message to be multicast as a single logical packet and as a sequence of physical packets, termed respectively as the Per-Message and Per-Packet approaches.…”

“…Next section will describe the overall architecture and the protocol of [1]; the former includes describing the roles played by the network monitoring and QoS negotiation components. The two extension approaches to multiple-packet RMcast will then be described in Section 3.…”