NCSU Libraries
Search the Collection|Browse Subjects|Services|Library Information|Community |News & Events

Title page for ETD etd-08062002-125435


Type of Document Master's Thesis
Author Adhikari, Akshay Arun,
Author's Email Address aaadhika@unity.ncsu.edu
URN etd-08062002-125435
Title Voice Over IP Performance Diagnosis
Degree Master of Science
Graduate Program Computer Networking
Advisory Committee
Advisor Name Title
Dr. Douglas S. Reeves Committee Chair
Dr Wushow Chou Committee Member
Dr. George N. Rouskas Committee Member
Keywords
  • SNMP
  • Voice over IP
  • self-similarity
  • clock synchronization
Date of Defense 2002-08-05
Availability unrestricted
Abstract
We investigate a framework for assessing the readiness of a network

to support Voice over IP (VoIP) at the pre-deployment

stage. In this framework, VoIP traffic is synthesized on the network, while

simultaneously monitoring the health of network devices and links using the

Simple Network Management Protocol(SNMP). Using this framework, we try to

understand whether SNMP can be used to detect which network links or devices,

if any, cause poor VoIP quality.

First, we investigate the limitations of the end-to-end VoIP quality

measurement, and SNMP measurement framework used in lab experiments. We

quantify the errors in end-to-end VoIP quality measurements, and in SNMP

measurements, so that these errors can be taken into account depending on

the application at hand.

Next, we use our lab experiments to understand how VoIP performance

metrics like delay and loss are affected by offered load on a link. From

our initial experiments with faulty synthetic traffic generators, we find

that even at low utilization, bursty network traffic can significantly

degrade VoIP quality, and small timescale measurements, which are

impractical with SNMP, are required to detect the problems. However,

using realistic emulation of network traffic in the lab, we find that when

network problems are severe and last for long periods of time, they can be

easily detected using SNMP. We also present a case study of VoIP

assessment data collected from a real network, where we again

successfully used SNMP to detect the network links that caused poor quality

Files
  Filename       Size       Approximate Download Time (Hours:Minutes:Seconds) 
 
 28.8 Modem   56K Modem   ISDN (64 Kb)   ISDN (128 Kb)   Higher-speed Access 
  etd.pdf 2.76 Mb 00:12:45 00:06:33 00:05:44 00:02:52 00:00:14