An ATM Call Model for Traffic Engineering
Project Award Date: 07-01-2000
The considerable success of traffic engineering for telephony networks is due in large part to the development of simple and statistically accurate call models. The goal of this project is to develop similar call models for Asynchronous Transfer Mode (ATM) networks, based on actual traffic measurements.
This project will make use of measurement capabilities to collect traffic data, to develop statistical call models from the data, and to show how these models can be useful for traffic engineering. Switched virtual circuits (SVCs) can collect call detail records (CDRs) that contain information about call start time, duration, ATM cell counts, ingress and egress points, and ATM service types and/or ATM adaptation layers (AALs). Analysis of these records will provide the necessary information to develop macro-level call models based on call arrival and duration statistics.
We will collect cell-level data at one or more network ingress or egress points (user-network interfaces, or UNIs) over a significant period of time and then correlate that data with CDR data to obtain statistics on activity within a given call. We will then develop micro-level call activity models to augment the macro-level call models previously obtained. Finally, we will use analytical and/or simulation techniques to demonstrate the usefulness of the developed ATM call models for traffic engineering activities such as link capacity planning.
Primary Sponsor(s): Sprint