Analysis of Dual-Bus Metropolitan Area Networks Using Distributed Quantitative Stochastic Simulation

Abstract

The Distributed Queue Dual Bus protocol (DQDB) has been adopted by IEEE and by the International Standard Organisation as a metropolitan area network standard. DQDB is also a European Telecommunication Standard. Using distributed simulation, this paper analyzes the performance of standard DQDB networks, DQDB networks with slot bandwidth reuse, as well as DQDB networks with slot preuse and reuse. The results suggest that although slot preuse is a conceptual mirror image of slot reuse, the maximum improvement in throughput of the networks, the causes of unfairness, and the stations which are unfairly favoured differ depending on the choice of bandwidth single-use, reuse, preuse, or pre-and-reuse. All quantitative performance measures were estimated to a specific precision using automated distributed stochastic simulation. The design of the DQDB simulators and practical implications of the results are also discussed.

A Parallel Simulation Methodology for Speedup and Obtaining Performance Estimates with Specific Accuracy: Experiences of its Application in Studies of Metropolitan Area Networks

“…The number of observations needed to estimate the access delay of segments transmitted from station i (as determined using SA-PTS) decreased strikingly as a function of i. DQDB is known to be unfair, e.g., see [24], and stations closer to the Head-of-Bus are less likely to be delayed from transmitting a segment. Consequently we intuitively expect that the sequences of access delays of stations near the Head-of-Bus are more positively correlated (since segments are less likely to be delayed by non-deterministic periods) than sequences from stations further downstream.…”

“…This DQDB's MAC mechanism can restrain a station from using empty slots for a nondeterministic period (duration of its COUNTDOWN state [24]), so that some empty ones are left for stations downstream. However, DQDB is known to be unfair, e.g., see [24], and stations closer to the Head-of-Bus are less likely to be restrained (have better access to empty slots). Also, they have more segments for transmission on that bus.…”

“…Already servicesl provided by networks using the DQDB protocol have been introduced in the United States, Europe, and Australia. Also, methods are under research for upgrading standard DQDB in the future; e.g., see [24]. The standard DQDB architecture and a full protocol definition are given elsewhere [15,24], and will only be outlined here.…”