Performance analysis of media access protocols for high-speed local and metropolitan area networks.

Abstract

Advances in optical fiber technology have opened up the possibility for networks capable of transmitting information at the rate of hundreds of megabits per second or even in the gigabit range. In a shared media network operating at such a high speed, the media access protocol is critical to performance. In this dissertation, we are concerned with modeling and analyzing the performance of two media access protocols for high speed local and metropolitan area networks, namely, the Fiber Distributed Data Interface (FDDI), and the Distributed Queue Dual Bus (DQDB) protocols. Our main performance measure of interest is the mean packet delay. Due to the high degree of interactions among stations in the networks employing these protocols, the exact analysis of these networks is extremely difficult. To simplify the analysis, we focus our attention on the models for a single station in the network. The effects of other stations are captured in the input parameters to the model. We present an analysis of M/G/1 queueing system with vacations in which the sojourn time of the server at the queue is controlled by a timer. The service time of a customer and the vacation time of the server are allowed to take on only integer values. With suitable timer-setting policy, our queueing system can be used as a model for a single station in the FDDI network which generates only asynchronous traffic at a single priority level. We obtain the distributions of the number of customers present in the system at an arbitrary point in time, the time the server spends at the queue in each cycle, and the number of customers at the beginning of a vacation. We analyze a discrete-time queueing system model of a single station in the DQDB networks. By imposing some constrained assumptions for analytical tractability, we are able to employ a numerical approach to obtain the distribution of the number of packets present in the system at an arbitrary point in time. From this distribution, the mean sojourn time of a packet can be computed.