REALTIME TWmSPORT PROTOCOL flP A WIRELESS ...

rockyboygangΔίκτυα και Επικοινωνίες

24 Οκτ 2013 (πριν από 3 χρόνια και 11 μήνες)

111 εμφανίσεις

REALTIME
TWmSPORT
PROTOCOL
flP
A
WIRELESS
EBvIR010a6EWT
Wan
Yi
Han
B.Eng.,
Xi'an
Jiaotong
University,
1989
M.
Eng.,
Nanyang
Tcchnological
University,
1996
THESIS
SUBMITTED
IN
PARTIAL
FULFILLMENT OF
THE
REQUIREMENTS
FOR
THE
DEGREE
OF
MASTER
OF
APPLIED
SCIENCE
In
the
School
of
Engineering
Science
Q
Wan
Yi
Han
2000
SIMON
FRASER
UNiVERSITY
December
2000
AU
rigtits
resemed.
This
work
may
not
be
reproduced
in
whole
or in
part,
by
photocopy
or
other
means,
without
permission
of
the
author.
uisitim
and
AcquWi
et
0'
iographic
Senrices
senrices
bibliographiques
"L9
The
author
has
granted
a
non-
exciusive
licence
allowing
the
Nationai
Lt'brary
of
Canada
to
reproduce,
loan,
distnbute
or
sel1
copies of this
thesis
in
microfonn,
paper or
electronic
formats.
The
author
retains
ownership
of the
copyright
in
this
thesis.
Neither
the
thesis
nor
substantial
extracts
fiom
it
may
be
printed
or
otherwise
reproduced
without
the
author's
permission,
L'auteur a accordé
une
licence non
exchisive
permettant
A
la
Bibliothèque
nationale du
Canada
de
reproduirea
pr&era
distri'buer
ou
vendre
des
copies de cette thèse
sous
la
forme
de
microfiche/film,
de
reproduction
sur
papier
ou
sur
format
électronique.
L'auteur
conserve
la
propriété du
droit
d'auteur
qui
protège cette thèse.
Ni
la
thèse
ni
des
extraits
substantiels
de
celIe-ci
ne
doivent
être
imprimés
ou
autrement
reproduits
saus
son
autorisation.
Red-time
Transport
EVotocol
(RTE')
is
a
widely
accepted
Intemet
Draft Standard for end-to-end transport of
red-the
data on the
Intemet,
such
as
IP
telephony and interactive
video.
This
thesis
project was an
experimental
study of the performance of red-time data transmission
in
a
wireless
LAN
test-bed
empioying
m.
It
aiso
established
an
OPNET
simulation
mode1
of
IiTP
in
a wireless
environment.
The
key
parameters
for
red-time
data transmission
stuâied
were
namely
delay, delay
jitter
and
packet
loss rate.
The
project
also
involved
waluating
the effects of
attenuation due to distance and
obstacles,
the
maximum
number
of
retransmissions,
the
dweU
üme,
local
network
traffic,
and
hard
handoff.
The results of the study demonstrate delay and
packet
loss of
sufflcient
magnitudes to interfere
with
usable
performance of
rd-time
communications.
Specificaily,
hard
handoffs,
background
LAN
traffic,
the
number of
allowed
retransmissions, the
direction
of data transfer, and
effects of
frequency
hopping,
all
were
shown
to
affkct
mal-time
performance of
our
wireless test-bed. The
experimental
study also
showed
anomalously
long
deiay
times
under
some
circulllstances
that
are
believed
to be
related
to
the
bel 1
time
in
the frequency hopping
spread
spectrum
LAN.
Dedication
To
my
parents
and
my
wife
1
would
Ue
to
thank
my senior supervisor,
Dr.
R
H.
Stephen
Hardy, for
his
guidance
and
support
throughout
the
duration
of my
research.
1
wili
always
remember
his
guidance, patience,
kindness,
and
encouragement.
1
wouid
also
Uke
to
thank
Dr.
Lji?jana
Trajkovic,
for
being on my
supenrlsory
committee
and
providing
access
to
her lab
utillties,
and
Dr.
Jacques
VaMy
for
bebg
my
examiner.
1
wouid
Uke
to
thank
BreezeCOM
for
providing
BreezeNET
PRO.
1 1
wireless
system
and
technical
support,
and
OPNET
Technologies
Inc.
for
providing
access
to
OPNET.
Table
of
Contents
..........................................................................................
Approvai
ii
Abstract
...........................................................................................
iii
.......................................................................................
Dedication
iv
............................................................................
Acknowledgments
v
............................................................
Abbreviations
And Notations
viü
....................................................................................
List
of Tables
x
List
ofngures
.......................................
........................................
..................................................................................
1
.
Introduction
1
............................................
1.1
Real-time
Transport
Protoc01
1
.................................................................
1.2
Thesis
ovenriew
6
.................................................
1.3 Contribution
of
the
thesis
7
II
.
Background
and
Related Work
....................................................
8
2.1 Related
RFCs
....................................................................
9
.....................................................
2.2 Additionai
related
work 11
.........................................
III
.
Experlment
of
RI F
Over
Wireless
LAN
14
3.1 Testbed and
testing
Plan
...................................................
16
.
3.1.1
Hardware
used
in
the
experiments
..................
.....
16
.............
3.1.2
Software
tools
used
in
the
measurements
17
........................
3.1.3
Generating
RTP
packet
stream
Ble
18
...........................
3.1.4
W
transmission
measurement
19
...................................................
3.2
BreezeNET
wireless
LAN
2 5
...........................................
3.3
Interference
considerations
2 7
3.4
Clock
compensation
..........................................................
29
.........
3.4.1
Sending
packets
fkom
wirelessl
to
wireless2
30
.........
3.4.2
Sending
packets
kom
wireless2
to
wireless
1
31
.......................................................................
3.5
Site
survey
3 3
.............................
3.6
R P
communications
without
handoffs
37
..........................................................
3.6.1
Packet
loss
3 7
..............................................................
3.6.2
Delay
4 0
.............................
3.6.3
Two
peak
aggregations
of
delay
42
...................................
3.6.4
The effect of network
trafRc
45
....................................
3.6.5
Delay
around
lost packets
46
.......................................
3.6.6
Lage
consecutive delay
5 0
.............................
3.6.7
The influence of the
dweli
tirne
51
..........................
3.6.8
The influence of retransmissions
53
........................................................................
3.7
Delay
jitter
82
3.7.1
The effect of
distance/locations
............................
82
...............
3.7.2
The effect of network
t dûc
and
air-llnk
85
..........
3.7.3
The effect of
dweii
time
and
retransmission.
88
..................................
3.8
RTP
communications
with
handoffs
91
.....................................................
IV
.
Discussions
and
Conclusions
118
......................................................................
4.1
Discussions
118
......................................................
........
4.2
Conclusions
...
127
..............................................................
Appendix
.
Simulation Work
130
...............................................................
A
.
1
Simulation
Tools
130
..................................................................
A2
The
RTP
mode1
131
A.2.1
The
network
mode1
...............................................
131
....................................................
A.2.2 The
node
mode1
132
.........................................
A.2.3
The
Ki?
process
mode1
136
A.3
Discussions
......................................................................
137
.......................................................................................
References
138
Abbreviations
And
Notations
AMR
Adaptive
Muiti-Rate
speech
codec
AP
Access
Point
ASB
Applied
Science
Building
ATM
Asynchronous
Transfer
Mode
CBR
Constant
Bit
Rate
CDMA/CA
Carrier
Sense
Multiple
Access
with
Collision
Avoidance
CDMA/CD
Carrier
Sense
Multiple
Access
with
Coiiision
Detection
CDV
Cell
Delay Variation
CSS
Center of System Science
FHSS
Frequency
Hopping
Spread
Spectnim
iESG
internet
Engineering
Steering
Group
IJ3TF
The
Internet
Engineering
Task
Force
MTC
International
Multimedia
Teleconferendng
Consortium
IP
Intemet
Protocol
ISM
Industrial,
Scientific,
Medical
band
LAN
Local
Area
Network
MAC Media Access
Control
MMUSIC
Multlparty
Multimedia Session Control
OPNET
Optimized
NMkrork
simulator
PLR
Packet
Loss
Rate
POTS
niblic
Old
Telephone
System
QoS
Quality
of
Service
RFC
Recommend
For
Comments
RSVP
&source
reSerVation
Protocol
RIT
Real-time
Transport Protocol
FUCP
Real-time Transport Control Protocol
RTSP
Real-Time
Streaming
Protocol
SA
Station
Adaptor
SIP
Session Initiation Protocol
TCP
Transport Control Protocol
UDP
User
Datagram
Protocol
VAT Video
Audio
Tool
VBR Variable Bit Rate
Mean
value
Standard
devfation
The propagation time of the
packet
The
packet
processing
time
by the
CPU
The
clock
àHerence
between
wireless2
and
wirelessl
at
time
Relative
clock
drifthg
rate between
wireless2
and
wirelessl
Tirne
at
wireless
1
Time
at
wireless2
List
of Tables
Table
3-
1
.
Site
survey.
Access Point
#
1
and Station Adaptor
.............
35
.............
Table
3-2
.
Site
survey.
Access Point
#2
and Station Adaptor
35
Table
3.3
.
The
packet
loss
iiequency
when wirelessl
(SA)
at
Merent
location.
AP
1
to
SA
.......................................................................
3 9
Table
3.4
.
The delay
around
some lost packets. data
tiom
Figure
3.25.
at
point
A
.........................................................................................
49
Table
3.5
.
Inter-packet delay for consecutive packets
with
large delay
50
.......................
Table
3-6
wirelessl
(SA) sending. at
different
location
84
Table
3-7
wireless2
(AP)
sending. wirelessl (SA) at
different
location
84
................
Table
3-8
Statistics
for delay
jitter.
wirelessl
(SA)
sending
86
Table
3-9
Statistics
for delay jitter.
wireless2
@Pl)
sending
..............
86
Table
3-10
Statistics for delay
jitter.
wirelessl
(SA)
sen-.
varying
dweil
time
and retransmission
.................................................................
9 0
Table
3-
1 1 Statistics for delay
jitter.
wfreless2
(AP
1)
sending.
m g
dweU
time
and retransmission
.........................................................
90
Table 3-12
Packet
loss
rate at handoff
............................................
9 2
Table 4-1
Packet
loss rates for the cases of the
number
of retransmission
............................................
to be
O
and
1
when
âwell
tirne
is
64
ms
127
Figure
1
.
1
Kï'P
data
packet
header
format
.......................................
5
...........................................
Figure
3-
1
.
The
creating
of
RIF
data file 19
......................................
Figure 3-2
Kï'P
transmissions
over
wireless
20
Figure
3-3
The locations of mobile station vs
.
base stations
..............
23
.................................................
Figure 3-4
An
isolated
wireless
I3\N
24
Figure 3-5 wirelessl and
wireless2
were
wire-hed
...........................
24
Figure 3-6 wirelessl and
wireless2
were
wire-lined
and not
connecting
to
........................................................................................
backbone
2 5
Figure 3.7
.
Radio signal
quality
for
AP1
and
SA
................................
36
...............................
Figure
3.8
.
Radio signal
quality
for
AP2
and
SA
-36
................................
Figure
3.9
.
Packet
lost rate at
different
location
38
Figure
3.10
.
Time difference
between
.
receiving time at
wireless2
and
....................................................
sending
time
at
wireless
1.
Point
A
55
Figure
3-
1 1
.
Time
Merence
between
receiving lime at wireless
1
and
sending
tirne
at
wireless2.
Point
A
....................................................
55
Figure
3-
12
.
Time
dttrerence
between
receiving time at
wireless2
and
.................................................
sending
time
at wirelessl. Point
B
5 6
Figure
3-13
.
Time
merence
between receiving
üme
at
wirelessl
and
....................................................
sending
time
at
wireless2.
Point
B
56
Figure 3-14
.
Time
difference between
receMng
time at
wireless2
and
....................................................
sending
time
at
wirelessl.
Point
C
57
Figure 3-15
.
Time
clifference
between receiving time at wirelessl
and
....................................................
sending
time
at
wireless2.
Point
C
57
Figure 3-16
.
Time
difference
between
receiving
time at
wireless2
and
..................................................
sending
time
at
wirelessl.
Point
D
5 8
Figure 3-17.
Time
Merence
between
receiving
time
at
wirelessl
and
sending time at
wireless2,
Point D
....................................................
58
Figure
3-18.
Time
ciifference
between receiving time at
wireless2
and
sending time at wirelessl, Point
E
. .. .
..............
.
...
...
..
.....
.. . . ......
.......
...
59
Figure
3-
19.
Thne
Merence
between
receiving
time
at wireless
1
and
sending time
at
wireless2,
Point
E
................
...................
........
.........
59
Figure 3-20.
Time
difference
between receiving time at
wireless2
and
sending
time
at wireless
1,
Point
F
.
. . ...... .
. . . .
.
. . .
. . . .
. . . . .
..
. . . . . . . . . . .
. . . .
.
. .
6 0
Figure
3-2 1.
Time
difference
between
receiving
tirne
at wireless
1
and
sending
time
at
wireless2,
Point F
. .
.
..
.....
.
..
.*..
.
. . .
. .
. .. . . . . . . . . . .
.
..
.
,
.
. . . . . .
6 0
Figure 3-22. Time
merence
between receiving
time
at
wireless2
and
sending time at wirelessl, Point
A,
zoomed
....
.
... . .
.. .
. . . .
.
..
.
. . . . . . .
..
. . . . . .
.
.
6
1
Figure 3-23. Time
difference
between
receiving time at wirelessl
and
sending time at
wireless2,
Point
A,
zoomed.
...............
..
.. .
.....
....
.. ..
6 1
Figure
3-24.Delay
of transmission
fIom
wireless 1 to
wireless2,
Point
A
62
Figure
3-25.Delay
of transmission
h m
wtreless2
to wireless
1,
Point
A
62
Figure
3-26.Delay
of transmission
h m
wirelessl to
wireless2,
Point
B
63
Figure
3-27.Delay
of transmission
fiom
wireless2
to wirelessl
,
Point
B
63
Figure
3-28.Delay
of
transmission
from
wireless
1 to
wireless2,
Point
C
64
Figure
3-29.Delay
of transmission
h m
wireless2
to wireless
1.
Point
C
64
Figure
3-30.Delay
of transmission
h m
wlrelessl
to
wireless2.
Point
D
65
Figure
3-3
1
.Delay of transmission
h m
wire1ess2
to wireless
1,
Point
D
65
Figure
3-32.Delay
of
transmission
from
wireless
1
to
wireless2,
Point
E
66
Figure
3-33.Delay
of transmission
h m
wire1ess2
to
wirelessl, Point
E
66
Figure
3-34.Delay
of transmission from wirelessl to
wireless2,
Point
F
67
Figure
3-35.Delay
of transmission
h m
wireless2
to wireless
1,
Point F
67
Figure
3-36.Delay
of transmission from wireless
1
to
wireless2.
Point
A.
Zoomed
.......................................................................................
6 8
Figure
3-37.Delay
of transmission
from
wireless2
to wireless
1.
Point
A.
............................................................................................
Zoomed
ôû
Figure
3.38
.
Delay for isolated wireless network.
wireless
1
sent
.......
69
Figure
3.39
.
Delay for
isolated
wireless network.
wireless2
sent
.......
69
Figure
3.40
.
Delay for isolated wireless network. wirelessl sent.
zoomed7O
Figure
3.41.
Delay for isolated wireless network.
wireless2
sent.
zoomed7O
Figure 3.42
.
Delay for wired network. wirelessl sent
........................
71
Figure 3.43
.
Delay for wired network.
wireless2
sent
........................
71
Figure 3.44
.
Delay for wired network. wireless
1
sent. zoomed
...........
72
Figure
3.45
.
Delay for
wired
network.
wireless2
sent. zoomed
...........
72
Figure 3.46
.
Delay for isolated wired
LAN.
wirelessl sent
.................
73
Figure 3.47
.
Delay for isolated wired
LAN.
wireless2
sent
.................
73
Figure 3.48
.
Delay for isolated wired
LAN.
wireless
1
sent. zoomed
....
74
Figure 3.49
.
Delay for isolated
wired
LAN.
wireless2
sent. zoomed
....
74
Figure 3.50
.
The
signaiing
message flow
during
hopping
..................
75
Figure
3-51
Delay when dwell time set to 64 ms.
wirelessl
sent. at Point B76
Figure
3-52
Delay
when
ciweii
time
set to
64
ms.wireless2
senkat
Point
B
...........
76
Figure
3-53
Delay when dwell time set to
64
ms.
wirelessl sent. at Point
B.
zoomed
..................................................................................
7 7
Figure 3-54 Delay when
dwell
time set to
64
ms.
wireless2
sent.
at
Point
B. zoomed
......................................................................................
77
Figure
3-55
Delay
when
retransmission set to
O.
wireless
1
sent. at Point
B
...........
78
Figure
3-56
Delay
when
retransmission set to
0.
wireless2
sent.
at
Point
878
Figure
3-57
Delay when retransmission set to
O.
wirelessl sent
.
at
Point
B.
zoomed
.................................................
79
2dii
Figure
3-58
Delay when retransmission set to 0,
wireless2
sent, at Point
B, zoomed
.........
....
. . ... . . . .
.
...
..
. .....
...
.....
..............
.
.....
. . .
....
.. . . .. ..
....
79
Figure 3-59 Delay when
dweii
time
set
back
to 32
ms,
retransmission
set
back to 1,
wirelessl
sent, at Point B
...........
.. .
.
.......
....
. .... . ... .
...
. . .
. .
.
8 0
Figure 3-60 Delay when
dweii
time
set back to 32
ms,
retransmission
set
back to
1,
wireless2
sent, at Point
B
....................
..
........................
80
Figure 3-61 Delay when
dwell
time set back to
32
ms,
retransniission
set
back to
1,
wirelessl sent, at Point B, zoomed
.......
......
....
.... .
...
.
.
.... .
...
8
1
Figure
3-62 Delay when
dweii
time
set back to 32
ms,
retransmission set
back to
1,
wireless2
sent, at Point B, zoomed
...................................
81
Figure
3-63
Delay density distribution,
wireless
1
(SA)
sending,
Point
A
94
Figure
3-64
Delay density distribution, wirelessl
(SAJ
sending, Point
B
94
Figure 3-65 Delay density distribution, wirelessl
(SA)
senàing,
Point
C
95
Figure 3-66 Delay density distribution,
wireless
1
(SA)
sending, Point D
95
Figure
3-67 Delay density distribution,
wirelessl
(SA)
sending,
Point
E
96
Figure
3-68
Delay density distribution, wirelessl
(SA)
sending,
Point
F
96
Figure 3-69 Delay density distribution,
wireless2
Wl)
sending,
wireiess
1
(SA)
at Point
A..
..
... . .. .. . . . . .
. .. .. . .
.
..
..
..
.
.. . .
..
.
. . . . . .
. . . .. . ... .. .. . . ..
.
. . . ..
....
... .. .
....
97
Figure 3-70 Delay density distribution,
wfreless2
(ml)
sendfng,
wirelessl
(SA)
at Point
B
............... ....... .............
................
. . . . . .
.
.
9 7
Figure 3-71 Delay density distribution,
wireless2
Wl)
sending,
wireiessl
(SA)
at Point
C
.......................................................
..........................
98
Figure 3-72 Delay density distribution,
wireless2
@Pl)
sending,
wirelessl
(SA)
at
Point D
....
. . . .. . .
.....
....
..
. .
..
......
.. .. .
.
..
.
.
.
.
. . .
. .
. .
.
98
Figure 3-73
DeIay
density distribution,
wireless2
W1)
sending,
witeless
1
(SA)
at Point
E.
..
. .
. . . . . . . . . . .
.
.
.
.
. .
.
...
. . . . . .
. . .
. . .
. . . . . . . . . . . . . . .
. . . . . . .
.
.
..
. .
.
.
. . .
. . .
9 9
Figure
3-74
Delay density distribution,
wireless2
(AP1)
sending, wirelessl
(SA)
at
Point F
.............................................................................
99
Figure
3-75
Delay density distribution, wireless
1
(SA)
sending,
Point
A,
zoomed
........................................................................
,
..................
100
Figure
3-76
Delay density distribution, wireless
1
(SA)
sending, Point
B,
zoomed
. . . . . . . .
.
. . . . . . . . .
.
. . . . . . . . . . . . . . . . .
. .
. . . . .
.
. . . . .
. .
. . . . . . . . . . .
.
. .
.
. .
. .
. . . . . . . . . .
.
. . . . .
.
. . . . .
1
00
Figure
3-77
Delay density distribution, wirelessl
(SA)
sending, Point C,
zoomed
.
. .
. .
.
.
. .
. . . . . . .
.
. . . .
.
. . . .
.
. . . . . .
.
.
.
.
. . . . .
. . .
. . . .
. .
. . . .
.
.
. . . . . .
. . . . .
. .
. . .
.
. . . . . . . .
.
. .
,
. .
. . . .
10 1
Figure
3-78
Delay density distribution,
wlreless
1
(SA)
sending, Point D,
zoomed
. . . . .
. . . .
.
. . . . .
.
.
. . . . .
. . . . .
. . . . .
. .
.
.
.
.
. . . . . .
. .
. . . .
.
.
. . . . . .
. . . . . .
.
. . . . . . .
.
. .
.
. . . . . .
. . . .
. . . .
. .
10
1
Figure
3-79 Delay density distribution,
wlrelessl
(SA)
sending,
Point
E,
zoomed
.............................
,.
.........
.
...........
.
...........
.
..........
.
102
Figure
3-80
Delay density distribution, wirelessl
(SA)
sending,
Point
F,
zoomed
..........................................
.
.............
.........
.
.......
. .
.
102
Figure
3-8
1
Delay density distribution,
wireless2
(AP
1) sending, wireless
1
(SA)
at
Point
A,
zoomed
.....
..
..
. . . .
. .
. . . . . . . .
. .
..
. . .
. .
. . . .
..
. . . .
. . . . .
. .
. .
. . . .
.
. . .
. . .
103
Figure
3-82
Delay density distribution,
wireless2
@Pl)
sending, wireless
l
(SA)
at
Point
B,
zoomed
...................................................................
103
Figure
3-83
Delay density distribution.
wireless2
@Pl)
sending, wirelessl
(SA)
at Point C, zoomed
.
.....
..
. . . .
.
. . . . .
..
..
. . .
..
. ..
...
. .
.. ..
. . . . . . .
.
. . . . . . . .
.
. . . . . . .
104
Figure
3-84
Delay density dfstribution,
wireless2
(AP1)
sending,
wirelessl
(SA)
at Point
D,
zoomed
...................................................................
104
Figure
3-85
Delay density distribution,
wireless2
(AP
1)
sending, wireless
1
(SA)
at
Point
E,
zoomed
......
.. .
......
...
...
..
...
....
. . .
. . ..
.
. .
. . .
.
. . .
.
.
. . .
.
.
105
Figure
3-86
Delay
demity
distribution,
wireless2
(Ml)
sending,
wire1essl
[SA)
at Point
F,
zoomed
.
.....
.. . .
. ..
..
....
...
. .
.....
. .
..
....
...
.
. . . . .
.
. .
105
Figure
3-87
Delay density distribution for wired network
connecting
to
backbone,
wireless
1
sending
.........................................................
106
Figure
3-88
Delay density distribution for wired network
connecting
to
............................................................
backbone,
wireless2
sencihg
1ûô
Figure 3-89 Delay density distribution for wired network connecting to
backbone, wirelessl
sencihg,
zoomed
..............................................
107
Figure 3-90 Delay density distribution for wired network connecting
to
backbone,
wireless2
sending,
zoomed
............................................
107
Figure
3-9
1
Delay density distribution for isolated wired network,
.............................................................................
wireless
1
sending
1
08
Figure
3-92 Delay density distribution for isolated wired network,
...........................................................................
wlreless2
sending..
108
Figure 3-93 Delay density distribution
for
isolated wired network,
...............................................................
wireless
1
sending,
zoomed
109
Figure 3-94 Delay density distribution for isolated
whd
network,
wireless2
sending,
zwmed
...........................................................
109
Figure
3-95 Delay density distribution for isolated wireless network,
...........................................................................
wireless
1
sending..
1
10
Figure
3-96
Delay density distribution for isolated wireless network,
.............................................................................
wireless2
sending
1 10
Figure
3-97
Delay density distribution for isolated wireless
network,
..............................................................
wireless
1
sending,
zoomecl
1
1
1
Figure
3-98 Delay density distribution for isolated wireless
network,
...............................................................
wireless2
sending,
zoomed
1
1
1
Figure
3-99
Delay density distribution
when
dweii
tirne
set
to
64
ms,
.................................................................
wireless
1
sent, at Point
B
1
12
Figure 3-100 Delay density distribution when dwell
iime
set to
64
ms,
wireless2
sent, at Point B
.....
..
.
. .. . . .
. . . . .
.
. . .. .
..
.
.
.
.
.
.
. . .
.
..
. . . . .
. .
.
.
.
.
. . . . . . . . . . . .
1
12
Figure 3-101 Delay density distribution when dwell
time
set to
64
ms,
wirelessl
sent, at Point
B,
zoomed
...................................................
113
Figure
3-102 Delay density distribution when dwell
time
set to
64
ms,
wireless2
sent, at Point
B,
zoomed
...................................................
113
Figure 3-103 Delay density distribution when retransmission set to
O,
wireless
1
sent, at Point B
....
.. .. .. .
......
. ..
....
....
.
....
.
......
.
. . . . . . .
1
14
Figure
3-
104 Delay density distribution when retransmission set to
O,
wireless2
sent, at Point B
.................................
..........................
. .
114
Figure
3-
105
Delay density distribution when retransmission
set
to
O,
wireless
1
sent, at Point B, zoomed.
.. ... . . . . . . . . .
.
.
.
..
. . . .
.
. . . .
.
. .
.
. . .
.
. . .
.
. . . . .
.
. . .
1
15
Figure
3-
106
Delay density distribution when retransmission set
to
O,
wireless2
sent, at Point
B,
zoomed..
.
. .
. . . . .
..
.
.
. .
. . . .
..
. . .
+.
. . . . .
. . .
.
. . .
.
. .
. .
.
.
1
15
Figure
3-107 Delay density distribution when dwell time set back to
32
ms,
retransmission set back to 1, wirelessl sent, at Point
B
.............
116
Figure
3-108 Delay density distribution when dwell
time
set back
to
32
ms,
retransmission set back to
1,
witeless2
sent, at Point
B
.............
116
Figure
3-109
Delay
density distribution when dwell time
set
back to
32
ms,
retrmsmfssion
set back
to
1,
wirelessl sent, at Point
B,
zoomeci
117
Figure
3-
110
Delay density distribution when
dwell
üme
set
back
to
32
ms,
retransmission set back to
1,
wireless2
sent, at Point
B,
zoomed
117
Figure
4-
1
The
percentage
of packets that exceeded 30 ms delay
... ...
122
Figure
4-2
The
percentage
of
packets that exceeded 100
rns
delay
....
122
Figure
4-3
The
percentage
of
packets that exceeded
300
ms
delay
....
123
Figure
4-4
The
percentage of packets that exceeded 30
ms
delay
together
with
lost packets
......,..
......
.
. ....
........
....
.
...
. .
. . . .
.
.
.
. . . .
. .
. . .
.
.
123
xvii
Figure
4-5
The
percentage
of
packets
that
exceeded
100
ms
defay
togethef
with
lost
packets
................................................................
124
Figure
4-6
The
percentage
of
packets
that
exceeded
300
ms
delay
together
with
lost
packets
...............................................................
124
Figure
A-
1
The
network
mode1
........................
.,
...............................
131
Figure
A-2
RTP
base
station
node
mode1
(transmit)
...........................
134
Figure
A-3
EUP
mobile station
node
mode1
(receive)
..........................
134
Figure
A-4
Jammer
node
mode1
.....................................................
136
Figure
A-5
RTP
process
mo
del
..........................................................
136
Chapter
1
Introduction
This
thesis
presents
rny
research
project in the
area
of
red-time
data communications
in
a
wireless
environment
with
the
implementation
of the
Real-time
Transport Protocol
m,
RFC1889)
[Schulzrinnel.
The
project consisted of the
experiments
on
W
wireless communications
and the development of
O P m
simulation models of the protocol,
focusing
attention on the performance of
RIF
over
urrireless.
The Real-
time
Transport Protocol was developed by the
Audio/Video
Transport
Working
Croup
of
Intemet
Engineering
Task Force
(IETF).
A
detailed
introduction to the protocol and
fts
possible
use
over
wireless is presented
in
this chapter
and
a
literature
review
is
provided
in
Chapter
II.
Chapter III presents experiments
together
with
the
correspondhg
results
while
Chapter
IV
presents discussions and
conclusions.
Kï P
is the Intemet-standard protocol for the transport of
real-time
data,
including
audio
and
video.
Interactive
senrices,
such
as
Intexnet
telephony and
Intemet
video,
as
well
as
media-on-demand
are
major
applfcations.
RTP
(RFC1889
where
RFC
stands
for
Recommend
For
Comments)
is
still
in
Intemet
Draft
Standard
status,
not
a
formal
standard.
It
was
first
proposed
in
November
1995
as
an
internet
proposed standard.
When
this
thesis
was
completed, the
draft
had
been
revised
several
times
and was
in
revision
8.
FtlF
consists of
two
parts: the data part
(KIT)
and the control part
W P ).
The data part
provides
support for
carrying
data
with
real-time
properties,
including
timing
reconstruction, loss detection,
security
and
content identification.
ni e
control part
(IITCP)
provides
support to
monitor
the quaiity of service and to convey information about the
participants
in
on-going
sessions
such
as
real-time
conferencing of
groups
of
any
size
on the
intemet.
The
support
includes
source
identiîlcation.
support for
gateways
such
as audio and video bridges and
also
multicast-to-unicast
translators.
It
provides
quality-of-service
(QoS)
feedback
from
receivers
to
the
rnulticast
group
and
also support for the
synchronlzation
of
different
media
strearns.
Some of the
RTP
use
scenarios
are
simple
muiticast
audio
conferencing, audio and
vldeo
conferenclag,
mixers
and
translators,
and
layered encoding.
RTP
can
make
use of the
IP
multicast
services of the
Internet
for
muiticast
voice communications.
If
a new participant
is
comected
through
a
low-bandwidth
Link
or
if
there
are
indications of network
congestion, senders
can
ahvays
change the encoding to
reduced
quaiity
during
communication.
Members
of the
group
that
are
cornmunicating
can
job
and leave
during
the
conference.
KKP
is
used for
collecting
the
data
such
as who
is
partfcipating
at
any
moment and how weli
they
are
receiving
the
audio
data.
When
both
audio and video media are used
in
a conference, they
are
transmitted
as
separate
E?TP
sessions.
KïP/#rCP
packets
are
traxismitted
for each medium
using
two
different
UDP
port pairs
and/or
multicast addresses.
With
the
separation
of audio and video media, some
participants
in
the conference
could
receive
only
one
medium
if
they
choose.
Some conference
users
may
be
connected
with
hi&-bandwidth
networks
while
other
are
comected
through
low-speed network access.
Instead
of forcing
everyone
to use a lower-bandwidth, reduced-quality
audiolvideo
encoding,
an
Kï'P-level
relay
calied
a
mixer
may
be placed
near
the
low-bandwidth
area,
translating
the
media
encoding to
a
lower-
bandwidth
and
forwarding
the
lower-bandwidth
packet
stream
across
the
low-speed
linEr.
Some
of the
intended
participants
in
the
audio/video
conference
might
not
be
directly
reachable
via
IP
multicast (for
example,
they
might
be
behind
an
application-level
firewall
that
WU
not let
any
IP
packets
pass).
in
this
case another
type
of
RîP-level
relay
called
a
tmnslator
may
be
used.
Wo
translators
are
instaüed,
one on
either
side
of
the
firewail,
with
the
outside
one
tunnelhg
ail
multfcast
packets
received
through
a
secure
comection
to the
translater
inside
the
firewall.
Another
functionality
of
KIT
is
that the multimedia
appiications
can
adjust
their
transmission rate
to
match
the
capacity
of the receiver
or to adapt to network congestion.
By
combining
a layered encoding
with
a layered transmission
system,
receivers
can
take
the
responsibility
for
rate-adaptation by using
KR.
RTP
also
provides
end-to-end
network
transport
functions
for
real-
tlme
data and
can
be
used over multicast or
unicast
network
semices.
Those
functions
include
payload
type
identification,
sequence
numberfng,
time-stamping
and
deiivery
monitoring.
KîP
usually
runs
on
top of
UDP
to
make
use of
Its
multiplexing
and
checksum
services.
Both
protocols
provide
parts of
the
transport
protoc01
functionality.
Efforts
have been made to
d e
RTP
transport-hdependent,
so that
it
may
be
3
used
with
other
suitable
underlying
networks or transport protocols
such
as
CLNP,
IPX
or
AAL5/ATM.
KTP
itself
does
not
provide
any
mechanism
to
ensure
timely
deihery
or
provide
other
quality-of-service
guarantees.
It
does
not
guarantee
delivery nor
does
it
assume that the
underlying
network
is
reliable and
deltvers
packets
in
sequence. Real-time delivery
always
requires
the support of lower
layers
that
actuaily
have
control
over
resources
in
switches and routers.
Error
recovery
mechanisms
(for
example,
recovering of lost packets) are not
detined
in
KP.
Such
mechanisms
are likely to be dependent on the
packet
content. For
example,
for audio, the
Eî F
has suggested
adding
low-bit-rate
redundancy
by
offsetting
in
time.
For other applications,
re-transmitüng
lost packets
may
be appropriate.. The
H.261
RîP
payload
definition
offers
such
a mechanism.
The
FkTP
data
part
has the header format as
shown
in
Figure
1-1.
The
total
length
of the header
is
96
bits
excluding
the
CSRC
identifier,
which
1s
used by
mixers
to
list
the
contributhg
sources of
the
mixed
RTP
stream.
With
timestamp,
the receiver
would
be able to put
the
incoming
audio
and video packets
in
the correct
timing
order.
With
the sequence
number, the receiver
can
detect
packet
loss and restore
packet
sequence.
The sequence number
increments
by one for each
K P
data
packet
sent,
timestamps
increase
by the time
"covered"
by
a
packet.
For video formats
where a video
fhme
is
split across
several
KTP
packets,
several
packets
may
have the
same
timestamp
but
different
sequence
numbers.
With the
SSRC
field the receiver identifies
the
synchronlzation
source. No
tm
synchronization
sources
within
the
same
W
session
will
have the
same
SSRC
identifier.
4
Synchronization
source
(SSRC)
identifier
(32
bits)
Contributing
source
(CSRC)
identiflem
(O
to
15
items,
32
bits
each)
V
Figure
1- 1
FUT
data
packet
header format.
FtiT
usually
runs
on top of
UDP,
not
TCP.
One
of the reasons is
that
reliable
transmission
fs
inappropriate
for delay-sensitive data
such
as
rd-tirne
audio
and
video.
RIF
delivers
real-time
data packets, which
would
not be
usefd
if
the delay were
too
long.
if
a
KI'P
packet
1s
lost or
P
meets
long delay when
it
arrjves
at the receiver,
it
might
be considered as
"lost"
since
it
might
be out of date
even
though
it manages to reach the
receiver.
Thus
TCP
is
not
suitable
for interactive
audio/video
which
Kï'P
is
focused on,
though
it
may
be
appropriate
for delivering audio and
video for playback.
TCP
may
often
nui
over
highiy
lossy
networks
with
acceptable
throughput,
where
the
uncompensated
losses
could
make
audio or video communications impossible. Some
other
reasons for
RTP
not
running
over
TCP
are
as
follows:
TCP
cannot
support multicast; the
TCP
congestion
control
mechanism
decreases the congestion window
when
packet
losses
are
detected, where the correct congestion response
for interactive media
is
to change the
audio/video
encoding,
the video
frame
rate, or the video
image
size
at
the
transmitter,
based, for
example,
on the feedback
received
through
m P
receiver
report packets.
X
CC
Payloadme
Sequence
Number
(16
bits)
For this project the
performance
of
the
E?ïT
in
a wireless
environment
was investigated.
As
Ftl'P
has
only
been tested
in
wire-lined
networks, how
RTP
performs
over
wireless
was
still
unknown
when this
project
began.
Experiments
in
this
thesis
were made to examine the
performance of
F t P
over wireless,
in
fked
locations and
in
handoff cases.
A
simulation
mode1
was
also
developed
to
simulate
ï?W
transmissions
over wireless.
in
these
experiments,
the
following
testing
scenarios
were
performed:
locate the
mobile
host at
digerent
distances/locations
to study the
performance at
Merent
received
signal
levels
establish a
wire-iined
local
area
network for the
sending
workstation and the
receiving
workstation to study the influence of
air-link
isolate
the base
station
fkom
the backbone to study
the
influence of
local network
t r a c
change the
dwell
time
of the wireless system to study
the
influence
of dwell
time
in
Frequency
Hopping
Spread
Spectrum
(FHSS)system
change the
maximum
number
of retransmissions
in
the wireless
system to study
the
iduence
of
retransmissions
perform
single
and
multiple
hard
handoffs
to study
their
influence
A
large
format
ffle
was
prepared
in
order
to undertake the
experiments.
1.3
Contributh
of
the
thes&
The
key
parameters
for
mal-time
data communications,
such
as
delay, delay
jitter
and
packet
loss rate, are
studied
in
the
irnplementation
of
K P
in
a
wireless environment
in
this
thesis
project. The factors of
background
LAN
trafBc,
the
number
of
aliowed
retransmissions, the
direction of data transmission, and the effect of
fiequency
hopping, were
all
shown
to
affect
real-time
performance of
our
wireless
RTP
test-bed.
The
results
of the
experiments
indicate
the large
packet
loss rate
and
some
packets
with
long delay
in
the wireless environment. The
packet
loss rate could be as
high
as
0.5%
and delay could be as long as
800
ms.
Both
the
large
packet
loss rate and long delay
make
KlT
unsuitable
in
the
wireless environment
studies
in
this
thesis.
The
wfreless
LAN
provided
by
BreezeCOM
used
hard
handoff,
which
is
not
acceptable for real-time data communication
using
F t P
because
the
packet
loss rate was too
high
-
at least
0.333%
for the single handoff
case, and
maximum
24.184%
when
four-time
handoffs occurred.
The
results suggest
that
hard
handoff not be used
in
real-time wireless
communication to avoid
enormous
packet
loss.
Chapter
II
BacWouad
anâ
Ralited
Work
With the explosive
growth
of the
Intemet
and
intranet,
a
great
deal
of attention has been
attracted
to the implementations and performances
of networked multimedia
semices,
which
involve
the transport of
real-
time
multimedia
streams
over
non-guaranteed
quaiity
of service
(QoS)
networks.
KIT,
the
Real-time
Transport
Protocol
(WC
1889)
[Schulzrinne,
Casner],
is
the proposed
internet
Standard
and has gained widespread
acceptance
as the transport
protocol
for
voice and
video
on the Internet.
RIT
was proposed by the
Audio/Video
Transport Working Group
under
The
internet
Engineering
Task
Force
(IETF).
The
internet
Engineering Task Force
(IETF)
is a large open international
communiîy
of
network designers, operators,
vendors,
and researchers concemed
with
the evolution of the
intemet
architecture
and
the
smooth
operation of the
intemet.
The
actual
technical
work of the
IETF
is
done
in its
working
groups, which are
organized
by
topic into
several
areas
(e.g.,
routing,
transport,
security,
etc.). The
Audioflideo
Transport
Working
Group
is
in the Transport
h a,
and
was
f o d
to
specify
a
protocol for
real-time
transmission of
audio
and
video
over
UDP
and
IP
multicast, and that
protocol
is
RTP.
The
process
of the stages
of
the
Internet
Standards is
Proposed
Standard,
Draft
Standard,
and
Standard.
At
the stages of
Roposed
Standard
and
Draft
Standard,
they
are
cded
Recommend
For
Comments
WC).
Not
every
RFC
is
an
Internet
Standard.
in
(Lennox]
the
author
concluded that
there
were
2149
RFCs
and
only
53
o£ficial
standards
then.
was
approved
by
The
Intemet
Engineering
Steering
Group
(IESC;)
as
an
internet
proposed standard on November
22,
1995.
Currently
K P
is
at
Draft
Standard stage.
RIT
has
been accepted
widely
as the transport protocol for voice
and
video
on the
Internet.
in
January
1996,
Netscape
announced that
their
Netscape
UveMedia.
a
standards-based
framework
for
bringing
real-time
audio
and
vide0
to the
Netscape
open software platform, would
be based on the Real-time Transport
Protocol
(m).
In
March
1996 The
international Multimedia Teleconferencing Consortium
(IhiTIYl),
which
is
the
group
pushing
for
an
open
Internet
communications
platfonn,
said
that its
hplementation
would be based on
iTü
and
IETF
speciflcations
including
RTP.
Microsoft
said
that
it
would include
these
communications capabilities in future releases of the
Windows
operating
system
and associated developer kits.
in
May 1996, Microsoft
claimed
that their
NetMeethg
Conferencing
Software supports
KIF.
iTü
study
gmup
(SG)
15
has agreed to use
KIF
for
LA.-based
conferencing
interoperable
with
H.320.
With the
proposa1
of
intemet
Standard
Ki?,
some
auxiliary
RFCs
and
internet
Drafts were proposed.
Accompanying
to
RIF
(RFC1889)
[Schulzrinne,
Casner],
RTP
Pro@?
for
Audio
and
Vicieo
Conferences
wWi
Mintnal
Control
(RFC
1890)
[Schulzrinne],
was
also
proposed for
Internet
Standard.
A
profle
cded
"~/A W
is
described
in
the document for the use of
the
real-time
transport
protocol
(RIF).
and the associated control protocol,
l?TCP,
within
audio
and video
multi-participant
conkences
with
minimal
control.
Interpretations
of
generic
fields
within
the
KïF'
spedcation
suitable
for
audio
and video
conferences
are described there. It
defines
a
set of
default
mappings
from
payload type
numbers
to
encoding,
and also
defines
a set of standard
encoding
and
the&
names
when used
within
KP.
In
general,
this
profile
defines
aspects of
RI F
left
unspecifled
in
the
RTP
protocol
definition
Some payload format
has
been
deflned
as
RFC
or
Intemet
Draft.
The
foilowing
is
a
iist
of
part
of
the
dehed
KîP
payload
formats:
MPEG-4
[Civanlar,
Casner]
,
AMR
[adaptive
multi-rate
speech
codec)
[Sjobergl
,
compressed
digital
video
data
streams
commonly
known
as
"DV'
(Kobayashi]
,
MIME
Type
Registration
[Casner,
Hoschka],
Cornfort
Noise
[Zopf),
MPEG-4
with
Flexible
Emr
Reslliency
[Guillemot, Christ,
Wesner,
Klemets]
,
MPEG-4
Audio/Visual
streams(Kikuchi],
12-bit
DAT,
20- and
24-bit
Linear
Sampled
Audio
[Kobayashi]
,
MP3
Audio
[Finlayson]
,
ITU-T
Recommendation
G.722.1
[Luthi],
SMPTE
292M
(for uncompressed
HiYïV)
[Gharai,
Goncher],
H.263
Video
Streams
[Zhu],
H.261
Video
Streams
mleffl],
Redundant
Audio
Data
[Perkins],
AC-3
(also
known
as
Dolby
Digital
or Dolby
AC-3)
[Gharai],
etc.
RI T
has
the
feature
of
scalability.
It
can
be
used for
unicast
and
multicast
h m
2
to
several
thousand.
Some
RFCs
and
Intemet
Drafts
have been
written
to
deflne
this
feature.
Different
Multiplexing
Schemes
or payload formats
can
be
found
in
mompson],
Bochmannl,
[Pazhyannur]
,
[Hoshi],
[Rosenbergl,
[Handey],
and
[Subbiah]
,
etc.
The
headers of
P,
UDP
and
RTP
together
contribute
40
bytes of
overhead to each
packet,
men
without
iP
options
and
RTP
CSRC
iists
or
header extensions.
Thus,
at
a
packetizatkon
internai
20
ms,
headers
alone
would
generate
16
kb/s,
too
much
for slow
serial
links.
In
order
to
reduce the
weight
of
header,
RIT
header compression has been
defined
10
in
the
following
RFCs
or
Intemet
Drafts:
[Koren],
[Bormann],
[Jonsson],
[Degermark]
,
[Casner and
Korenl,
[Casner and
Jacobsen],
etc.
Packet
loss
is
always a problem for multimedia
conferencing.
The
disruption
caused
by
even low
loss
rates
may
convince
a whole
generation of
users
that multimedia conferencing over the
Internet
is
not
viable. The
addition
of
redundancy
to
the
data
stream
is
offered as a
solution.
If
a
packet
is
lost
then
the
missing
information
may
be
reconstructed
at the receiver
ikom
the
redundant
data that arrives in the
following packets. Some
other
techniques
except
redundancy
include
retransmission,
interleaving
and
forward
error
correction. The error-
fighting
methods
were
defiaed
in
[Perkins]
,
[Perkins, Kouvelas]
,
[Perkins,
Hodson],
[Rosenberg],
Dl,
[Nguyen],
[Guillemot],
[Civanlar,
Cash,
Haskel], and
ITU's
proposal
bng].
There
is
another
protocol,
cailed
Red-TIme
StreQming
Protocol
(KïSP,
RFC
2326)
[Schulzrinne,
Raoj.
This
protocol
allows
controlling
multimedia
streams
deiivered,
for
example,
via
RIF.
It
is
mainiy
used for
Internet
media-on-demand.
SiP,
the Session
Initiation
EVotocol
(Handley,
Schulzrinne],
is
a
signahg
protocol for
intemet
conferencing, telephony,
presence,
events
notitication
and instant
messaging.
SiP
was
developed
within
the
Ei'F
MMUSIC
(Multiparty
Multimedia Session
Control)
working
group,
with
work
proceeding
since
September
1999
in the
IETF
SiP
working
group.
Some
means
of
combating
with
packet
loss or error bits were
defined
by
some
RFCs
or
Internet
Drafts
discussed
in
previous
section.
They
include redundancy, retransmission,
interleaving
and
fomard
error
correction, etc.
If
the real-time
traî3c
encounters
traf3c
jam,
packet
delay and
delay
jitter
would make
these
late-arrfval
packets useless, and the
above-
mentioned
methods
would
be
not
helpfui
for
reducing
delay and delay
jitter. Resource
ReSerVation
Protocol
(RSVP,
RFC
2205)
[Braden],
is
part
of a
larger
effort to
enhance
the
current
Internet
architecture
with
support for
Quaiity
of
Semice
(QoS)
flows.
ni e
RSVP
protocol
is
used by a
host to request
specac
qualittes
of
sexvice
from
the
network
for
particular
appiîcation
data
streams
or flows.
RSVP
is also used by
routers
to
deiiver
quality-of-service
(QoS)
requests to
al1
nodes
dong
the
path(s)
of the flows
and
to
estabiish
and
maintain
state to
provide
the
requested service.
RSVP
requests
will
generally
result
in
resources
being
reserved in each
node
dong
the
data
path.
While
as
RSVP
tries
hard
to
reserve the
resource,
it
may
not get
what
it
requested
since
the
node
may
not have
sufacient
available
resources.
A
primary
feature
of
RSVP
is
its
scalability.
RSVP
scales to
very
large multicast
groups
because
it
uses
receiver-oriented
reservation
requests
that
merge as they progress up the
multicast tree. The
reservation
for
a single
receiver
does
not need to
travel
to
the
source of
a
multicast tree;
rather
it
travels
only
unüi
it
reaches
a reserved
branch
of the tree.
While
the
RSVP
protocol
is
designed
speciûcaily
for multicast
appiications,
it
may
also
make
unicast
reservations.
As
RSVP
provides
a
mean
for
offering
predictable
QoS,
The Session
Description
Protocol
(SDP,
RFC
2323
[Handley,
Jacobson]
is
another
assist
mechanism,
and
offers
the
ability
to
the
basic
signahg
protocol to
describe
call
session
parameters.
It
is
used
to
assign
session
parameters
to
every user
participating
in
a point-to-point or
multipoint
call.
ITU-T
recommends
H.323
IIlU-q
for packet-based multimedia
communication.
This
recommendation
is
the
pioneering
overall
speciflcation
for
implementing
packet-based multimedia
conferencing
over local
area
networks
that
cannot
guarantee
quaiity
of service.
Many
protocols
make
up
the
complete
H.323
standard
group,
to
provide
signaling
and
c d
processing.
Henning
SchulPinne
[Schulzrinne,
website]
also mentioned that
KîP
is
a member of
H.323
family.
An
objective
measuement
of
the
voice
quality
was
specified
in
ïïü-
T
Recommendation
P.861.
It
specifies
a
mode1
to
map
actual
audio
signals
to
their
representations
inside
the head of a
human.
Details
can
be found
in
the
Recommendatlon.
Kl'P
was
modeled
in
Ethemet
using
OPNEï
in
[Mlnton].
A
simulation and
periormance
evaluation
of
KïP
was used
in
their
experiment
course. The
modehg
was for the
teaching
purpose.
Slmilar
work was also
done
by
[Ko].
While
the
characteristics
of
wire-iined
networks
have
been
well
documented, less
experimental
data
is
available
for
wjreless
LANs.
In
[Eckhardt]
the
authors
reporteci
the
results
of
a
study
characteridng
the
wireless
error
environment
-
in-building
wireless
network.
They
evaluated the effects of
interfering
radiation sources
and
of
attenuation
due to distance
and
obstacles on the
packet
loss rate
and
bit
error rate.
The
authors
found
that
under
many
conditions the error rate of
this
physical layer is
comparab1e
to
that
of
wired
links.
Chapter
ïïï
Expriment
of
RTP
Oret
Wirelebs
LAN
The
purpose
of the
experiments
was
to evaluate the performance of
the Real-time Transport
Rotocol
(KP)
in
a
wireless
environment.
As
K P
is
used for the transmission of
media
with
rd-time
property,
such
as
voice and
video,
delay
and
deiay
jitter
are
major
concems.
Thus
reiiable
connection between the
sender
and
the
receiver,
which
needs the
re-
transmission of packets on
the
transport layer,
is
not necessary.
That
I s
why
RTP
runs
on
UDP,
not
T'P.
To evaluate the performance
of
RTE'
in
wlreless
LAN,
the
following
measurements
were
performed:
Site
survey.
With
a
fùred
location of the base station (Access Point),
when the location of mobile station [Station Adaptor) varies,
the
distance and obstacles between
them
Vary,
also
the received signal
quality
Çom
each
other
varies. Site
survey
was
used to assess
the
received signal
strength
of the
point-to-point
link
Clock
compensation. The
docks
of the computers
In
use were
different
from
the
absolute
time
and
always
drif'üng
ikom
the
absolute
time.
What
important
to the
experiments
is
not the
absolute
time
ciifference
and
drifthg
rate between
individual
computer
and
absolute
time,
but
the
relative
thne
ciifference
and
relative clock
drifting
rate between the
receiving
and
sending
computers.
The
delay of
the
transmission
should
be
compensateci
by the relative
Merence
and relative
dock
drifting
rate;
The
RIF
transmission
between
base station and mobile station
when mobile station changes
its
location;
The
EtTP
transmission between mobile and base stations when
they
established a
stand-alone
wireless
WUIJ
The
RIF
transmission
between
the
sending
and
receiving
workstations
when they estabiish a
wire-itned
LAN,
with
connection to the backbone
The transmission between
the
sending
and receiving
workstations when
they
establish a stand-alone
wire-hed
LAN
The
RIT
transmission between the mobile and base stations when
the mobile station
performs
handoff
In
order
to
perform
the
above
experiments, a testbed
is
established
and iiiustrated
in
Section
3.1.
Also
testing
methodology
is
discussed
there.
ni e
wireless
system
used
in
this
experiment
is
BreezeCom's
BreezeNET
PRO.
1 1
(BreezeComl.
It uses Frequency Hopping Spread
Spectrum
(FHSS)
technology
in
2.4GHz
-
2.835GHz
ISM
(industrial,
Scientific,
and
Medical)
band and the
maximum
transmission rate
is
3
Mbps,
which
is
among
the
highest
systems
operated
in
the
ISM
band.
The
details
are described
in
Section
3.2.
Radio
signais
transmitted
îkom
a base station
are
not
only
subject
to the
same
signiflcant
propagation
losses,
but are
also
subject to other
interferences.
The interferences on the
air-link
in the
experiments
are
discussed
in
Section
3.3.
Delay should be the receiving
üme
minus
the
sen*
time,
if
they
use the
same
dock.
The
dock
drifting
of the receiver and the
sender
makes
the calculation of delay
quite
complex.
Not
on@
at the
start
point
of a transmission the sending computer and receiving computer have
dock
Merence,
but also
during
the transmission session, the
clock
15
ciifference
varies.
A
compensation
method
for clock
merence
and
clock
drifting
for the
sender
and the
receiver
is
dexived
in
Section 3.4.
Site
s we y
was
perfonned
to
assess
the radio signal
quality
of a
point-to-point
link
between the
Access
Point,
which
connected
to
the
ResearchNet
LAN,
and
the Station Adaptor,
which
connected to the
mobile host. The
radio
signal
quality
is
discussed in Section 3.5.
The
results
of
W
transmissions with
various
locations of mobile
station
are
discussed
in
Section
3.6.
Packet
loss rates, delay and
also
the
innuence
of
network
kaflic
and
air-link
are
discussed
there.
in
Section
3.7
we
talked
about
delay
jitter,
another
important
measure
besides delay for communication over internet.
When the handoff happens, how
does
the
KïP
transmission
perfonn?
In
Section
3.8
we
have a
brief
view.
3.1
Testbed
and
testing
Pian
In
order
to
perform
the
desired
experiments,
a
testbed was
estabiished.
3.1.1
Hardware
useà
in
the
experhmt s
The
testbed
was
consisted
by
1
wired
10/
100BaseT
Ethemet
LAN
-
ResearchNet
in
Center of
System Science
[CSS)
in
Simon Fraser University;
2
identicai
Deil
Dimension
XPS
PCs
with
PII
333
CPU,
64
MB
memory, and 3COM
3c509
Ethemet
adapter each. One
was
named
as
wire1essl.ensc.sfit.ca,
P
address
199.60.6.61;
another
was
named
as
wireless2.ensc.sfu.ca,
IP
address
199.60.6.62;
16
2
identical
Sun
ültra
10
workstations. One
was
named
as
bacon.easc.sfu,ca,
IP
address
199.60.6.1
17;
another
was
named
as
carver.ensc.sfit.ca,
IP
address
199.6û.6.118:
1
set of
BreezeNET
PRO.
1
1
wireless
system,
including
2
Access
Points
(AP),
1
Station Adaptor
(SA);
1
FM
radio receiver, used to generate audio input.
The
setup
of the testbed was
shown
in
Figure
3-2.
The
Sun
workstations
ran
on
Sun
Solaris
2.6.
The
PCs
ran
on
Redhat
Linm
6.0
with
kemel version
2.0.33.
Linux
is
Unk
iike
operating
system.
It
is open source and
can
be
freeiy
distributed.
The software
version for
BreezeNET
PRO.
1 1
was version
3.4.
3.1.2
Softulare
took
used
h
the
measurements
The software
tools
used were
RTP
toolset
developed
by
Henning
Schulainne
and the version was
1.12
[SchulPinne,
websitej.
The
tools
used
in
our
experiments
include:
rtpplay
-
playback
RIF
sessions (or
can
be a
dumped
KïF
file)
rtpdump
-
iistens
to the
address
and port for
K P
and
RTCP
packets and
dumps
a processed version to stdout or a file
VAT
-
a
real-time
multimedia audio
conferencing
application over
the
internet.
Based on the
Intemet
Draft
Standard
Real-time
Transport
Protocol
(RTP),
it
was
developed by the
Network
Research
Group of Lawrence Berkeley National
Laboratory.
The
version
was
4.Ob
1
WG].
Modification
was
made by us to the source files of
KîP
tools
to
increase
the
precision
of the
receiving
time and the
sending
time
fiom
17
miilisecond
level to microsecond level. The
tools
were re-compiled
d e r
modification.
The basic
measurement
procedure
is:
at the sending
side,
use
rtpplay to play a pre-recorded
KlP
file,
sending
the
RTP
packet
stream to
a
specified
network
address
with
a
port
number. The
sender's
information,
such
as
packet
sending
time,
packet
sequence number,
timestarnp,
source id, etc
is
recorded
in
a sending
repoft;
at the
receiver
side,
the
corresponding
information,
such
as
packet
receiving
time,
packet
sequence number, etc
is
also
recorded
fn
a
receiving
report,
using
rtpdump.
With
the
above
information
the
delay, delay
jitter
and
lost
packets
can
be
calculated.
3.1.3
Cenerattng
RTP
packet
streamjb
A
RIF
format
BLe
was
created
fkst
for
playtng
with
rtpplay.
The
ille
was recorded by using the
dump
format of
rtpdump.
Since
VAT
is
based
on
the
Internet
Draft
Standard
RTP,
it
was
used to generate the
EiTP
packet
stream. The creation of
the
Kï'P
file
was
iliustrated
in
Figure
3-
1.
FM radio receiver
carver.ensc.sfu.ca
bacon.ensc.sfu.ca
199.60.6.118 199.60.6.1 17
Figure
3-1.
The
creating
of
RTP
data file.
While
VAT
was
running
on
caruer,
the
audio
signal
fkom
the
FM
radio
receiver went
into
the
me-in
port of
carver.
The
input
in
VAT
was
changed
h m
M C
to
Line-in,
so
that
the
VAT
could
generate
R'P
packet
stream
and
send
this
stream
to
bacon.
At
the
same
time
rtpdump
was
running
on
bacon.
It
dumped
a
flle
accoràing
to the
paclret
stream
it
received.
The
fite
was
KIF
format
and
can
be played
using
rtpplay.
The radio station
whkh
the
radio receiver
tuned
to was
Praise
FM 106.5
in
Vancouver.
The
ECïP
Ne
was
name
as
radio3.rtp.
There
were
12,587
KiP
packets
in
the
fie,
total
playing
üme
was
about 8 minutes
23
seconds.
The
measurement
for
KW
transmission
over
the
wireless
LAN
was
performed.
The setup of
the
testbed
was
Uustrated
in
Figure
3-2.
wireless
1
.ensc.sfu.ca
199.60.8.61
APl
AP3.
Figure
3-2
K F
transmissions over wireless.
Wireless
1
together
wl th
Station Adaptor
(SA)
is
the mobile station.
Access Point
(AP)
becomes the
base
station. The locations of
APs
are
ked.
With
the variation of the
mobiie
station's location,
the
received
signal
strength
at
the
receiver
also varies.
Thus
the performance of the
KIF
transmissions also varies.
The
first
set of
experiments
is
the
study
of
KIF
performance
with
the mobile station
in
different
locations. The locations of the mobile
station and the base station are
shown
in Figure
3-3.
There
are
two
base
stations,
AP
1
and
AP2.
MI
is
located at Point
A.
AP2
is located at Point
G.
Mobile station
would
be located at Point
A,
B.
C,
D,
E,
F,
G
and
H
respectfvdy
to
perfonn
the
experiments.
To
study
the effect
of
local network
traffic
on the delay of
RïT
transmissions,
a
second set of experiments are
perfonned.
By
isolating
AP
1
and
wireless2
from
the backbone
(ResearchNet),
an
independent
wireless network
is
thus
established,
as shown
in
Figure
3-4.
The
comection
is
APl
and
wireless2
to
hub,
and
SA
to wirelessl.
In
this
case, wireless
1
and
wireless2
can
communicate
by
air-llnk,
without
the
interference of the local network
tdTic.
The influence of
air-link
on
EirP
transmission
is
examhed
in
the
third
set of experiments.
By
comecting
wireless
1
directly
to the
backbone as shown
in
Figure
3-5,
wirelessl and
wireless2
becoxne
two
wl re-hd
nodes
in the
same
LAN.
EZTP
transmissions
are
performed
here.
In
this
case, the
air-link
factor
did
not
exist
any
more.
Cornparison
experiments for
RTP
transmissions on the
wire-lined
LAN
without
local network
traffic
are
performed
in the
fourth
set
of
experiments. The
expriment
setup
is
shown
in
Figure
3-6.
The
packets
on the
air-link
are
prone
to be
compted
or lost.
In
RIF,
mtransmission
in
transport
layer
does
not
function.
In
Daka
Link
Layer
of the wireless system, the wireless system
itself
has
a
setang
for
retransmission.
in
the
BreezeNET
PRO.
1
1, by default the
maximum
number of retransmission
is
set to
1.
That
is to
Say,
if
a
packet
is
lost or
compted, the
sender
(either
AP
or
SA)
will
retransmit the
packet.
If
again
this
packet
is
lost or compted,
it
won't
be
transmitted
and the
receiver
will
nwer
receive
it.
The user
is
told that this
is
opümal
setting.
What
could
happen
if
there
were no retransmissions?
By
setting
the
parameter
of
maximum
number
of retransmission to
0,
the
Mth
set of
experiments,
wbich
is
with
the backbone as shown
in
Figure
3-2,
and
the
sixth
set
of
experiments.