notes04 - TELCOM 2120 Comparing Systems via Measurements...

Info iconThis preview shows pages 1–4. Sign up to view the full content.

View Full Document Right Arrow Icon
1 TELCOM 2120 Comparing Systems via Measurements and Regression Joseph Kabara Department of Information Science and Telecommunications University of Pittsburgh Spring 2008, notes4 TELCOM 2120: Network Performance 2 Four basic types of measurement studies Four basic types of measurement studies 1. Statistical Characterization of metric – Determine: call blocking at telephone switch, utilization of network links, bit error rate on link, etc. 2. Compare alternate system designs, equipment or parameter settings – Benchmark routers from different vendors, compare different token rotation times in FDDI ring, compare RAS, etc. Types of Measurement Studies Types of Measurement Studies Spring 2008, notes4 TELCOM 2120: Network Performance 3 3. Predicting System Performance 3. Predicting System Performance – Throughput and Delay increase if add 5 workstations to an Ethernet switch. – End-to-end delay of virtual private network link in ISP backbone networks. 4. Parameterise analytical or simulation model or driving a simulation model – Determining mean call holding times of phone calls for use in queueing model. – Gather data to use as traffic source in computer simulation model of alternate network topologies. Types of Measurement Studies Types of Measurement Studies
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
2 Spring 2008, notes4 TELCOM 2120: Network Performance 4 • How does one compare data from two random samples drawn from different populations? Sample X Sample Y Population X Population Y Statistical Inference Statistical Inference Comparison of Two Populations Comparison of Two Populations Spring 2008, notes4 TELCOM 2120: Network Performance 5 For example Compare the mean performance of two different routers from different vendors. Compare the mean opinion scores of two voice compression techniques. Can compare based on either confidence intervals or hypothesis test. Simplest approach is based on confidence interval comparisons – applies to any parameter (mean, probability, quantile, etc. .) Comparison of Two Populations Spring 2008, notes4 TELCOM 2120: Network Performance 6 Confidence interval comparisons on the parameter q x and q y from populations X and Y respectively Construct a 100(1- a 1 )% confidence interval ( a,b ) on q x and a 100(1 - a 2 )% confidence interval ( c,d ) on q y. One then compares the confidence intervals 1. Confidence intervals don’t overlap a > d q x > q y with confidence level 100(1- a 1 /2 - a 2 /2)% 2. Confidence intervals don’t overlap c > b q x < q y with confidence level 100(1- a 1 /2 - a 2 /2)% 3. Confidence intervals overlap can’t conclude any difference between q x and q y with confidence level 100(1- a 1 /2 - a 2 /2)% Comparison of Two Population Parameters
Background image of page 2
3 Spring 2008, notes4 TELCOM 2120: Network Performance 7 The form of the confidence intervals will depend on the population parameter to be compared from populations X and Y.
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Image of page 4
This is the end of the preview. Sign up to access the rest of the document.

This document was uploaded on 11/02/2009.

Page1 / 26

notes04 - TELCOM 2120 Comparing Systems via Measurements...

This preview shows document pages 1 - 4. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online