Tunable reliable multicast for periodic information dissemination

Abstract

Existing reliable transport protocols for periodic information dissemination ignore application semantics while attempting to be 100% reliable. "Application Level Framing" (ALF) suggests that taking application semantics into account when designing transport protocols can result in performance that is highly optimized for the network. We apply this principle in designing a policy-based TUNAble semi-reliable multicast protocol (TUNA) for periodic information dissemination. Specifically, TUNA is not constrained to guarantee full reliability, but allows the receiving application to selectively request retransmissions of lost portions of the data stream, based on user-level policies. TUNA uses statistical properties of the data stream to adaptively guide receivers in dynamically altering their reliability policies. This is particularly well matched for satellite systems, where end node to satellite bandwidth is limited, especially when shared by large community of end nodes. Our simulation results show that TUNA is made highly scalable by reducing the number of NACKs sent back to the information source, while keeping the staleness of received data within application-specified bounds. Under large session sizes, TUNA yields a factor of 1.5 to 6 improvement over a fully reliable scheme; for high loss rates, it results in a factor of 1.5 to 5 improvement. TUNA is also able to reduce contention for retransmission request bandwidth in a multiple access backchannel environment. We have implemented TUNA and a prototype application called InfoCaster in the MASH toolkit.