A partial-destination-release strategy for the multi-token ring protocol

Abstract

A strategy for ring networks using multiple tokens is defined. The tokens are released either at their sources or at their destination nodes, depending on bandwidth availability and frame lengths. This results in greater bandwidth availability and reduced access delays, as confirmed by simulation results. The protocol is shown to outperform several other popular protocols, like the token ring with early token release, e.g. FDDI, the slotted ring, and DQDB. This performance superiority is shown to hold over almost the entire range of traffic loads, with different network spans and different values of average packet lengths.<>