The design and optimisation of protocols is fundamental to operation of the Internet. The history of protocols follows a S-curve, similar to other high-tech products: New products require significant levels of fundamental research as an understating of the key issues grows, followed by a combination of research and development (engineering) as actual problems are solved, and finally by a commercial development phase in which there are relatively few technical problems to be addressed.
Development of TCP (the unicast transport protocol in the Internet) followed such a curve over the past 20 years and is now considered by most communications engineers to be mature. It is to be expected that TCP will continue in its current form for the foreseeable future, until it is replaced by an entirely new protocol.
For multicast, the current situation is very different. Much of the existing understanding of design issues learned from TCP no longer apply. Over a decade of basic research has been completed, researchers are now at the point where they are reaching consensus on the key issues and the first two generation of multicast products have emerged.
There is now a need to define new multicast protocols which address the problems that have been identified. This may be done by practical experiments to observe actual behaviour together with research to refine protocol techniques. This will, in turn, lead to a new generation of protocols. Work is specifically needed to improve the scaling to large networks, reliability, and specific issues concerning integration into next generation satellite networks.