SLIDE 1
Wireless
Traditional – node is one hop away
from access point
Ad-hoc
no access point multiple hops are the norm nodes theselves forward packets.
People come and go as they please
Collaborate on the fly(cool term!)
AODVoice Voice over AODV Durga Prasad Pandey 6.829 Course Project - - PowerPoint PPT Presentation
AODVoice Voice over AODV Durga Prasad Pandey 6.829 Course Project - - PowerPoint PPT Presentation
AODVoice Voice over AODV Durga Prasad Pandey 6.829 Course Project Fall 2006 Wireless Traditional node is one hop away from access point Ad-hoc no access point multiple hops are the norm nodes theselves forward
SLIDE 2
SLIDE 3
AODV
One of the most popular ad-hoc
wireless routing protocols
Reactive - Only when you desire Deletes non-used nodes from RT ~ 10
seconds
AODV-UU
SLIDE 4
Motivation?
Why Voice?
Heavy enough to stress the wireless
network and reveal weaknesses
Time sensitive Video can buffer, sensors don’t send much
data
Its an application!
SLIDE 5
Have a compelling need? ;)
5000000 Children’s Laptops (OLPC) Mesh networking enabled Huge potential for on the fly voice
conferencing, playing!
Testbed and Usebed.
SLIDE 6
What did we do?
A C B Call between A and C, sometimes goes through B
SLIDE 7
Approach
Play an audio file over a SIP phone
repeatedly and see how well it is received
Initial approach:
Play sound file at A’s speaker Capture it with microphone Send it over linphone Play linphone output on speaker Capture with microphone and record
SLIDE 8
How dumb!
SLIDE 9
Refined approach
Use Jack: outputs audio from one application
to another application, very low latency!
At A:
alsaplayer o/p linphone i/p
At C:
linphone o/p rezound i/p
What these are:
Alsaplayer =jack capable music player Linphone = SIP phone Rezound = sound recorder/player/mixer
SLIDE 10
Where did this happen?
SLIDE 11
Specifically!
SLIDE 12
Which song have you heard 500 times in the last week?
SLIDE 13
Measuring voice quality
ITUT recommendation PESQ: Perceptual Evaluation
- f Sound Quality
Mean opinion score(MOS)
Scale 0 = worst to 4.5 = best 2+ = comprehensible
Why not human ear?
Not an objective, consistent instrument
Considered most accurate speech quality standard Measures effects of dustortions such as delay, noise
etc to model and predict subjective quality
SLIDE 14
Results – One hop fixed
One Hop MOS variation with time
0.5 1 1.5 2 2.5 3 3.5 4 4.5 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 Time PESQ MOS
SLIDE 15
Effect of using Jack
One HOP MOS variation for fixed position
0.5 1 1.5 2 2.5 3 3.5 4 4.5 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 Time PESQ MOS Series1 Series2
SLIDE 16
Results – 1 hop moving
A verage PESQ MOS variation with distance for 1-hop
0.5 1 1.5 2 2.5 3 3.5 4 4.5 1 2 3 4 5 6 7 8 9 Distance PESQ MOS
SLIDE 17
Results: 1 to 2 hops, moving
MOS variation when number of hops changes
0.5 1 1.5 2 2.5 3 3.5 4 4.5 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Distance MOS Score
SLIDE 18
Results: Hop oscillations
MOS Oscillation at Hop threshold
0.5 1 1.5 2 2.5 3 3.5 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Time PESQ MOS
SLIDE 19
Observations
Ping not reliable? Obstructions more lethal than distance +ve control signals might not imply
data is reaching destination
Assymetric signalling If routing is stable, 2 hop voice not bad Wireless signals can vary a lot!
SLIDE 20
Future Work
More frequent hellos Configure(hardwire) for 2/more hops
and measure MOS
Try ETX Measure buffering delays, which might
become significant with many hops
SLIDE 21
Thanks!
Special thanks to Dina, Rob and Sachin
for excellent suggestions
To Kwan for asking: Can’t you connect