Dissemination-Oriented Communication Systems.

Abstract : This is the final report for ARPA contract no. DABT63-91-K-0001. This contract was focused on exploring dissemination oriented communication systems, systems in which the primary mode of communication was based on publishing information sources multicasting to subscribing sources. The approach was develop new protocols at the network, transport, session and presentation layers, and evaluate these protocol designs by implementation and use with a few sample dissemination applications that are also being developed. The objective was to design a next generation communication protocol architecture for dissemination oriented communication that was applicable in a wide variety of (inter)networks, ranging from high speed, high delay terrestrial to low speed, high error rate mobile networks. We started the contract with some basic ideas of how to revise the various protocol levels to support the dissemination model, based on our prior experience with IP multicast, VMTFP and application-layer protocols. Over time, these ideas were refined such that some were incorporated into a common protocol, others were discarded as unworkable or suboptimal. and some were realized as originally proposed. It would probably be politic to claim we are highly successful in every aspect of the original proposed work. However, usually some of the greatest insights and progress come from mistakes, and in that vein, we had lots of insights and progress. In this report, we also try to be honest and clear about how some of our original proposed directions failed, and why. The following sections describe the results for each of the original deliverables, plus sections corresponding to what we actually developed, if that was different from originally proposed. We first describe some background ideas and how their development affected our work on this contract.