A comparison of two token-passing bus protocols

摘要

A well known disadvantage of standard token-passing in ring and bus networks is the waste of channel bandwidth often seen in lightly loaded or asymmetric systems. It is possible to make use of the broadcast mechanism in token bus systems to distribute nearly up-to-date information about the state of individual stations to the entire system. One such scheme involves the determination of a randomly varying set of more active stations. These stations are given a chance to form a second logical ring above the standard logical ring that characterizes the token bus. The transmission cycles of the system can thus be made to alternate between standard token-passing and transmission cycles, and the cycles of token-passing and transmission within the logical ring of more active stations. We assume that each station makes at most one transmission when given the chance to transmit. For Poisson arrivals and otherwise general input distributions, the cycle-time distribution of the token is derived for each kind of cycle. An important random variable is the random token turnaround time seen by individual stations. For lightly loaded stations this time tends to be larger than for heavily loaded stations. The distribution of this random time, simple performance measures, and a comparative measure of stability, showing the adaptive scheme to be more stable than the standard, are obtained.