Semester

Spring

Date of Graduation

2006

Document Type

Thesis

Degree Type

MS

College

Statler College of Engineering and Mineral Resources

Department

Lane Department of Computer Science and Electrical Engineering

Committee Chair

Arun A Ross

Abstract

Computer networks are connected with other networks using one or more dedicated circuits. These dedicated circuits have a finite amount of available bandwidth to transport the data. There is a need to predict the future utilization of these circuits so that additional capacity can be added before the circuit becomes saturated. Once the circuit becomes saturated, network packets will be discarded resulting in a poor end-user experience. It typically takes several weeks or, in some cases, several months for an order of additional capacity to be installed. A network planner should, therefore, know in advance when additional capacity will be required. The goal of this thesis is to develop a system that can predict the traffic utilization of a circuit six months into the future. If this goal can be accomplished, then network planners will have the ability to optimize the provisioning cycle of network capacity. The thesis describes a method to build a model for estimating network traffic data in the future based on current network characteristics. It also proposes a technique for anomaly detection that can be used to determine if the model has to be updated as traffic characteristics may change over time. The performance of the model has been evaluated on six different real-world datasets. Experimental results indicate the strengths and limitations of the proposed model.

Share

COinS