ECE6610 VoIP Final Presentation

8/15/2019 ECE6610 VoIP Final Presentation

1/22

VoIP Call Capacity

Group 2

Team Members: Blake Brannon

Steve ConoverGuillaume-Jean Herbiet

Balasubramanian Seshasayee


2/22

ECE 6610 202/09/08

Introduction

Problem Overview

Topology and Results Summary

Attempted Improvements

Conclusions


3/22

ECE 6610 302/09/08

Problem Statement

Using open source tools, setup

appropriate network topologies and

measure VoIP call capacity for 802.11band 802.11g networks

Develop solutions to improve call capacity


4/22

ECE 6610 402/09/08

Motivation

WLANs are widely deployed

VoIP has a promising future as the

dominant voice communication method

WLAN ad-hoc networks are ideal for

communicating in disaster areas

Important to measure the quality/capacity

of VoIP traffic over WLANs


5/22

ECE 6610 502/09/08

Milestones

Team Submits

Requests and

Team Profile

Final

Presentation

Interim

Report

802.11b

Two Hop

Dr. Siva

Presents

Projects

Initial

Metrics

Recorded

802.11g

Two Hop

802.11b/g

Infrastructure

T topology Improve

More

Metrics

Recorded


6/22

ECE 6610 602/09/08

Packet Loss and R-Factor

0

10

20

30

40

50

60

70

80

90

100

0 10 20 30 40 50 60

Packet Loss (%)

R-Factor

R-FactorITU Recommendation G107 Introduced E-model

which outputs R-factor based on perceptual and

equipment impairments in its calculation.

Very Satisfied

Satisfied

Some Users SatisfiedMany Users Dissatisfied

Nearly All Users Dissatisfied

Not Recommended


7/22

ECE 6610 702/09/08

Topologies Overview

Infrastructure

Two Hop

Cross Traffic T

Two Hop with Improvements


8/22

ECE 6610 802/09/08

Infrastructure

Software: Linux, kPhone, iPerf, Wireshark

Mix of lab and personal hardware Recording on both kPhone clients!

`

Wired System

iPerf ServerkPhone

Wireshark

192.168.1.10

Access

Point

Router

192.168.1.1

iPerf

192.168.1.12

kPhone

Wireshark

192.168.1.11


9/22

ECE 6610 902/09/08

Theoretical Node Throughput

0 10 20 30 40 500

500

1000

1500

2000

2500

3000

3500

4000802.11 Single Hop (No RTS/CTS)

Number of Nodes (n)

No

de

Throug

hput(Kbps

)

802.11b

128kbps VoIP Requirement

802.11g


10/22

ECE 6610 1002/09/08

Infrastructure (Cont.)

348Analytical

42286Theoretical Max

3011Measured

802.11g802.11bCalls Supported

Supported calls is substantially lower than expected Small packets and packet overhead reduce maximum

calls Values inflated due to limited number of computers in the

testbed


11/22

ECE 6610 1102/09/08

Loss Rates


12/22

ECE 6610 1202/09/08

Jitter


13/22

ECE 6610 1302/09/08

Linear Multi-hop

Adhoc setup with Linux system acting as relay Increased contention Wireshark recording on endpoints

(NOT in the middle!)

Relay

192.168.1.13 kPhoneiPerf ClientWireshark

192.168.1.11

Wired System

iPerf ServerkPhone

Wireshark

192.168.1.10


14/22

ECE 6610 1402/09/08

Linear Multi-hop (Cont.)

Multiple hops reduces maximum

Higher contention Excessive delay (2-5 seconds mean delay in test

cases)

93CallsSupported

802.11g802.11b


15/22

ECE 6610 1502/09/08

Cross Traffic T-Mode

Increased contention

Start of a grid

VoIP

CrossTraffic

RelayiPerf Server192.168.1.13

kPhoneiPerf ClientWireshark

192.168.1.11

Wired SystemiPerf Server

kPhone

Wireshark

192.168.1.10

iPerf Client192.168.1.14


16/22

ECE 6610 1602/09/08

T-Mode (Cont.)

Computers become overloaded in relay

-5CallsSupported

802.11g802.11b


17/22

ECE 6610 1702/09/08

Improvements

Turn RTS/CTS on

VoIP packet size increase

Packet Aggregation

Relay192.168.1.13

kPhone

iPerf ClientWireshark

192.168.1.11

Wired System

iPerf ServerkPhone

Wireshark

192.168.1.10


18/22

ECE 6610 1802/09/08

Improvements (Cont.)

RTS/CTS introduced extremely large delay but no

improvements

Some improvement in increased packet size

Aggregation indicated small improvement

Small improvements hard to measure in environment

5

Double

Size

3-53CallsSupported

AggregateRTS/CTS


19/22

ECE 6610 1902/09/08

Conclusions

WLAN characteristics result in large delay

and low VoIP call capacity

Ad-hoc networks more affected by delay,infrastructure more affected by loss

Improvements can be made by reducing

the number of packets contending at anygiven time


20/22

ECE 6610 2002/09/08

Thanks for your attention

Feel free to ask questions!


21/22

ECE 6610 2102/09/08


22/22

ECE 6610 2202/09/08

All Losses

ECE6610 VoIP Final Presentation

Documents

Transcript of ECE6610 VoIP Final Presentation