Documente Academic
Documente Profesional
Documente Cultură
I.
INTRODUCTION
17
9th IEEE International Workshop on Performance and Management of Wireless and Mobile Networks, 2013
Delay (msec)
ITU
CISCO
Conversational based services
3GPP
ITU
Videoconferencing
Services
ITU
CISCO
3GPP
PingER
3
1
3
2.5
CISCO
3GPP
ITU-T
150
150-400
10 seconds
CISCO
5 seconds
Date rate
QoS indicators
Highly
Interactive
Lectures/Discussions
(Videoconferencing)
16-384 Kbps
Interactive
Lectures/Discussions
(Voice applciations)
4-320 Kbps
16-384 Kpbs
Packet loss
Non-interactive
Lectures (Streaming)
15-400
REQUIREMENTS
APPLICATIONS
ITU
CISCO
3GPP
Study
based on a
developing
country [7]
ITU
CISCO
150-400
Services
150-400
150
5
1
5
18
9th IEEE International Workshop on Performance and Management of Wireless and Mobile Networks, 2013
19
9th IEEE International Workshop on Performance and Management of Wireless and Mobile Networks, 2013
H.263
MPEG-4
Metric
Value
QP
Resolution
QCIF (177*144)
tmn encoder (Version
2.0/3.2)
16 kbit/s
ON
3603.28 sec
19436
Single Layer
MOMUSYS MPEG-4
QCIF 176x144
89998
7499
12,
IBBPBBPBBP
BB
5
Encoder
Bit rate
USE of PB-frames
Length
No. Frames
Layer
Encoder
Frame Size
No. Frames
Number of GoP
INTRA PERIOD
QP
Layers
Application
RLC
Parameters
Voice calls
Streaming
video
Videoconfe
-rencing
Mode
Timer
MRW
(msec)
Timer
Discard
(msec)
MAX
MRW
MAX DAT
QoS class
Conversational
Payload size
224 bits
Background
Variant
Conversational
576 bytes
Unacknowledged
900
7500
6
4
DCH
PHY
Channel
type
Conversational
Bit
TTI
rate
(msec)
(Kbps)
Background
Bit rate
(Kbps)
TTI
(msec)
64
144
20
10
20
9th IEEE International Workshop on Performance and Management of Wireless and Mobile Networks, 2013
Packet loss
(%)
Li
Ui
Delay (msec)
Li
Ui
Voice application
150
400
Streaming
application
Videoconferencing
application
N/A
N/A
150
400
Delay
Figure 4. Comparison of packet loss of voice application after adding a
streaming client with different codec
Number of
streaming
client
H.263
10
12
14
16
18
20
5.58
1.82
5.56
1.85
5.75
1.89
5.93
1.89
5.79
1.91
5.62
1.94
5.75
2.02
Packet loss
(%)
13.55
Streaming application
1.0287
Videoconferencing
application
4.02
0.4
0.8
Weight
Videoconferencing
Voice
Streaming
Videoconferencing
0.4
0.6
0.2
0.6
value. Our work has been inspired by the QoS analysis method
proposed in [6]. The authors in their work have used two
performance points of an application for QoS metrics under
consideration. They have used these points to calculate the QoS
of an application on the fly. First, they calculate the normalized
value of each QoS metric using the performance metric of the
serving network. Then they provide a weight for each QoS
metric according to its relevant importance.
In our case, the application performance measurement is
collected from the average value of a certain QoS parameter
after the application is run for a certain period of time in the
considered network. These measured values are used with the
acceptable QoS parameter values which are defined earlier to
calculate the application QoS value. Then this application QoS
value is used to measure the network QoS value. Although, in
this work the network QoS is evaluated engaging one network
technology, the aim of this scheme is to involve different
network technologies which will be stated in our future work.
This scheme helps to find the best available combination in
context of networking technologies, applications and number
of users.
This approach also unfolds a way for bridging network QoS
with Quality of Experience (QoE). User satisfactions of a
network are closely related to the application performance. For
instance, a voice based service with a lower delay, normally
results in higher user satisfactions. So, if the application
performs as the user expects the user usually gives a better
satisfaction rating for the network. From the analysis of section
II, it is apparent that each service has its own QoS
requirements. To find out the performance of a particular
network, the performances of application-based traffic flows
present in that network are analyzed. We define the lower and
the upper bound values of QoS metrics for each service from
the analysis in the previous section. Table VIII illustrates these
values in the context of packet loss and delay. The bound for
packet loss of voice application is from 1% to 3%. For
streaming based service, the range is from 1% to 5% and for
Application
Voice
Streaming
21
9th IEEE International Workshop on Performance and Management of Wireless and Mobile Networks, 2013
Delay
(msec)
Packet
loss (%)
md,n m pl ,n
8
10
14
18
216
224
203
215
2.06
2.92
2.94
3.23
0.74
0.70
0.79
0.74
QoSn
0.47
0.04
0.03
0
0.63
0.50
0.48
0.44
QoS n = mi ,n wi
i =1
lower bound values for the specific QoS metric and the average
for the same QoS metric value derived from each application
flow in the network are put into Equation (1) to calculate the
normalization value for each QoS metric.
mi,n=
(3)
Umi
Lmi
Umi
0,
1,
Umi Lmi
(2)
QoS = QoSn
n =1
(4)
Delay (msec)
md,n
QoSn
m pl ,n
QoS
Voice
Streaming
Voice
Streaming
Voice
Streaming
Voice
Streaming
Voice
Streaming
Voice
Streaming
202
176
2.92
1.82
0.79
0.04
0.49
1.49
10
221
179
2.95
1.85
0.72
0.025
0.43
1.43
14
208
180
4.41
1.89
0.79
0.47
1.47
18
219
180
6.46
1.94
0.72
0.43
1.43
22
9th IEEE International Workshop on Performance and Management of Wireless and Mobile Networks, 2013
REFERENCES
[1]
ACKNOWLEDGMENT
We would like to thank OPNET for providing us with
Modeler software license.
23