Analysis of delay characteristics in intelligent networks

Abstract

This dissertation is concerned with the following four issues: 1) how to analyze delay characteristics in queueing networks with infinite buffers, 2) how to analyze delay characteristics in queueing networks with finite buffers, 3) how to analyze intelligent networks (IN) based on the SS7 signaling network, and 4) how to analyze IN based on ATM networks. Telecommunication networks can be modeled as open queueing networks consisting of a number of nodes. We propose an approximation method for deriving mean sojourn time and sojourn time distributions in queueing networks with infinite and finite buffers. The approximation method is based on decomposition, a two-moment approximation, a PH/PH/1 waiting time approximation, and convolution of PH-type distributions. The approximation method is validated by comparing the approximation results with simulation results in case of traffic condition (low or high traffic), four different types of queueing networks consisting of 9 nodes with different servers (deterministic, Erlang, exponential or hyperexponential servers), and buffer size (infinite or finite). IN concept is to provide complex services over a less sophisticated signaling system using intelligent servers, called the service control points (SCPs) that extend the capabilities of the network. We introduce an IN/ISDN signaling platform and a protocol mode, and evaluate the performance of the IN/ISDN signaling platform in terms of mean sojourn time and sojourn time distribution using the approximation method. It becomes very important and necessary to obtain such distributions, especially for quantiles in the tail distributions with the advent of advanced networking technologies. The ATM-based B-ISDN has been recognized as a target network which provides sufficient bandwidths for transfer of all types of information, efficient mechanisms for switching and multiplexing, and enhanced service control capabilities. The IN-based signaling has been chosen by sev...