Protocol design and modeling issues in broadband networks

The authors present a new paradigm which suitably takes into account the underlying network environment of high bandwidth, greater availability of storage, and low error and loss probabilities, as well as multicast applications. Specifically, they describe a simple and efficient error detection and recovery protocol for multicast communication, which is an enhanced version of the pure negative acknowledgement scheme. As an illustration, they discuss error and flow control procedures that are functionally simple and hardware realizable, and offer good delay throughput performance. The high bandwidth, fixed cell sizes, and regularities in the data and signaling traffic suggest that conventional equilibrium models must be complemented by transient models which capture the dynamic behaviour of the system.<<ETX>>