PREDICTION OF VIDEO QUALITY OVER IEEE802.11
WIRELESS NETWORKS UNDER SATURATION CONDITION
Xijie Liu and Tarek N. Saadawi
Electrical Engineering Department, City University of New York, City College, New York, U.S.A.
Keywords: Video quality, Markov Chain, IEEE 802.11.
Abstract: In IEEE 802.11 wireless channel, and under the assumption of ideal channel, packets are lost when it
exceeds the maximum retry attempt. Such packet losses lead to degradation in the transmitted video quality.
This paper provides an analytical approach to estimate the video quality distortion due to the packet losses
in IEEE 802.11 wireless networks. The analytical approach is based on the use of two-state Markov chain
model combined with the ITU-T Recommendation G.1070 for video quality objective measurements. Our
approach provides a relationship between the design parameters of IEEE 802.11 wireless channels and the
required video quality.
1 INTRODUCTION
Video transmission over IEEE 802.11 wireless
networks has become a popular application in
wireless communication. Hence, evaluating the
video quality video quality over IEEE 802.11
wireless networks has drawn much attention. Khan
et al provides, reference (Khan, 2009), provides
quality prediction for various video content types
over wireless networks. Methods to estimate the
distortion due to packet losses in wireless video
communication are provided in references (Babich,
2008), (Bouazizi, 2004) and (Choi, 2005).
In IEEE 802.11 wireless networks, a packet will
be discarded when it exceeds the maximum packet
retry limit in IEEE 802.11 protocol. This packet loss
results in an inherent distortion of video quality,
even if the network operates in an ideal physical
environment.
Quality of Experience (QoE) has been addressed
in ITU-T G.1070. QoE grades the perceptual video
quality by mapping subjective quality to peak signal-
to-noise ratio (PSNR). The opinion model of ITU-T
G.1070, (correlating objective video quality with its
subjective video quality), provides a tool to estimate
the video quality thus measuring user's specific level
satisfaction. The model is able to eventually map the
packet loss ratio of the transmission channel to the
quality degradation and predict the video quality.
The goal of this paper is to determine a
relationship between the design parameters of IEEE
802.11 wireless network and the required video
quality. The basic idea of our approach is determine
an analytical expression for the packet loss rate
using the Markov chain model developed in
(Bianchi, 2000). Once this packet loss rate is
determined, we make use of the video quality
distortion formula in ITU-T G.1070. With this
approach, we can calibrate various IEEE 802.11
parameters to control the video quality distortion and
thus meet the required level of video quality. The
rest of the paper is organized as follows. Section 2
provides the analytical video quality in IEEE 802.11.
Section 2.1 is a summary of the opinion model of
ITU-T G.1070, while section 2.2 discusses analytical
packet drop rate with Markov Chain Model. Section
2.3 provides the numerical analysis results for video
quality in IEEE 802.11. Section 3 is the conclusion.
2 ANALYTICAL VIDEO
QUALITY IN IEEE 802.11
WIRELESS NETWORKS
2.1 Opinion Model of ITU-T
Recommendation G.1070
In ITU-T G.1070, video quality parameters are
introduced, such as video delay, T
[ms], video
packet-loss rate, P
, and jitter. These parameters
affect video quality when video is transmitted over
127
Liu X. and N. Saadawi T..
PREDICTION OF VIDEO QUALITY OVER IEEE802.11 WIRELESS NETWORKS UNDER SATURATION CONDITION.
DOI: 10.5220/0003448601270130
In Proceedings of the 6th International Conference on Software and Database Technologies (ICSOFT-2011), pages 127-130
ISBN: 978-989-8425-76-8
Copyright
c
2011 SCITEPRESS (Science and Technology Publications, Lda.)
wireless networks. According to ITU-T J.241, an
input value of T
must be less than hundreds of
milliseconds and a jitter less than tenths of
milliseconds in order to be tolerated for high quality
video streaming services. Video packet-loss rate
(P
) refers to end-to-end video packet-loss rate and
should be less than 10 %.
ITU-T G.1070 provides an algorithm that
estimates video quality. According to the ITU-T
G.1070, objective measurement of video quality, V
,
is calculated by:
=1+
−
(1)
where D
is degree of video quality robustness
due to packet loss; P
is video packet-loss rate;
I
represents the basic video quality affected by
the coding distortion under a combination of video
bit rate and video frame rate. I
is objective
measurement of basic video quality accounting for
coding distortion.
Every video content has its own video quality
robustness D
, and its own objective
measurement of basic video quality accounting for
coding distortion, I
. These two values are able
to be derived by applying the method described in
ITU-T G.1070. We only cite some measurement
from (You, 2009), and list them in Table 1(To
simply discussion, we assume D
is constant
under different P
and I
is constant with
different video bit rate and video frame rate). Thus,
in this paper, we only focus on how to obtain video
packet-loss rate, P
, using Markov chain analytical
model (Bianchi, 2000).
Table 1: Coefficients of I
coding
and D
PplV
for a video.
value 3.655 0.0037
2.2 Analysis Model under Saturation
Condition
We use the two-dimensional saturation Markov
chain models shown in Figure 1 to analyze the
packet dropping rate of IEEE 802.11. In the analysis,
we assume that the wireless networks operate in an
ideal physical environment, being the same one in
Bianchi’s model (Bianchi, 2000).
In the discrete-time Markov Chain shown in Figure
1, we define b
,
as the stationary distribution
probability of being in state (j,k), where j∈
0,L
)
is
the backoff stage, k∈0,w
−1 is the backoff
counter and w
is the contention window size at
backoff stage j.
Define m as maximum backoff stage when
contention windows will double. By the Markov
Chain regularities, a normalization requirement,
1=
∑∑
b
,
+
∑
b
,,
, and w
=
2
w
j≤m
2
w
m< ≤
,
we obtain
1
b
,
=
1−2p
2
1−p
)
−
p−p
2
1−p
)
1−p
)
+
w
2
1−p
)
1+
2p −
2p
)
1−2p
)
+
2
p
−p
)
1−p
(2)
where p is a probability that a node senses the
channel busy in a random slot. We denote τ the
transmission probability that a node attempts to
transmit a packet in a randomly chosen slot time.
Knowing that any transmission occurs when the
backoff time counter equals to zero, we will have
Equation (3).
τ=
b
,
=
1−p
1−p
×b
,
(3)
Substituting equation (2) into equation (3)
furthermore, we obtain equation (4) for the node’s
transmission probability τ.
τ=
)
)
)
)
)
)
)
(4)
Equations (4) are called the IEEE 802.11 node
property formula since it represents a binary
exponential backoff scheme to access to the
medium. It determines the node's transmission
probability in terms of the channel busy probability
as well as the network configuration parameters
(L,m,w
). The set of variables of
p,τ
in equation
(4) will be regarded as the attributes of a
transmission of an IEEE 802.11-based station with
arbitrary traffic arrival rate. Noticing that every node
i will have its own
p,τ
, we now attach the node’s
serial number to
p,τ
and formulae (4) become
equation (5), where i=1,2,… N; and N is number of
nodes in the network.
(5)
τ
=If the packet has not been successfully
transmitted after packet retry limit L times
attempting, the packet is dropped. Hence, the packet
dropping probability can be estimated as: p
=
p
(collision L+1 times before dropping). If the
ICSOFT 2011 - 6th International Conference on Software and Data Technologies
128
traffic is video, (and assuming IP and TCP/UDP
does not increase packet loss rate) then the video
packet-loss rate is obtained by
P
=p
(6)
Probability, p
in equation (5), that a node senses
the channel busy in a random slot, depends on the
transmission status of its neighbors nodes and varies
from one node to the other. If given a network
topology, we can obtain the other equation for every
node i’s p
. Then using numerical solutions, we are
able to solve equation (5), and ultimately obtain
every node’s objective measurement of video
quality, V
by equation (1).
Figure 1: Markov Chain model for saturation.
For example, we discuss video quality in a single
hop wireless network under saturation condition.
Assuming a wireless network has N nodes; all nodes
are in a single hot coverage area. Hence we obtain
p=1−
1−τ
)
(7)
We can solve equations (5) and (7) numerically.
The parameters L,m,w
,N will determine, then
affect V
. If parameters L,m,w
,N change, then V
will change; we will compare V
with IEEE
802.11a/b/g in our numerical calculations (Table 2).
IEEE 802.11a and IEEE 802.11g should have the
same results.
Table 2: System default parameters /configuration.
802.11a 802.11b 802.11g
CWmin 15 31 15
CWmax 1023 1023 1023
L 6 5 6
M 6 5 6
2.3 Numerical Results of Video
Quality,
In order to understand what are the optimal setting
parameters for IEEE 802.11 network to achieve
certain video quality, we change the packet retry
limit, L, minimum contending windows, w
, and the
number of nodes, N. We also let m=L. We notice
that the packet retry limit L has more effect on the
video quality than the minimum contending
windows, w
, as seen in figure 2. When L=5, V
changes more moderately with the increase in the
number of nodes in the network. On the other hand,
when L=2, V
decline rapidly with the increase in
the number of nodes; only with network size of
N=2, V
>4 can be achieved.
Figure 2: IEEE 802.11 setting parameters to achieve
certain video quality V
.
Figure 3: Video packet loss rate (under saturation
condition and basic access mode).
However, as we notice in Figure 3, video packet-
loss rate (P
) exceeds 10% when L=2,w
∈
27,31
)
andN∈
20,48
)
), which exceeds the
requirement of ITU-T G.1070 and ITU-T J. 241.
For optimal 802.11 parameter IEEE settings we
p−1
p−1
p−1
p−1
p−1
p−1
p
2, −
m
Wm
1
W
p
0,m
1, −
m
Wm
0,L
1,L
2,L
2, −
L
WL
1, −
L
WL
2,0
0
−W
1,0
0
−W
2,0 −
j
W
1,0 −
j
W
j
W
p
m
W
p
0,1−m
L
W
p
0,0
0,1−j
0,j
1,j
2,j
1,02,0
1,
m
2,m
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
)1( p−
)1( p−
)1( p−
)1( p−
)1( p−
)1( p−)1( p−
)1( p−
)1( p−
)1( p−)1( p−
)1( p−
)1( p−
)1( p−
)1( p−
5 10 15 20 25
1
1.5
2
2.5
3
3.5
4
4.5
5
number of nodes
Vq
w
0
=20
w
0
= 31
L=2
L=5
L=3
w
0
:20 to 31
L=4
10 20 30 40 50
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
number of nodes
drop rate
w
0
=31
L=3
w
0
=27
L=2
L=5
L=4
w
0
=27
w
0
=27:31
w
0
=27
w
0
=27
PREDICTION OF VIDEO QUALITY OVER IEEE802.11 WIRELESS NETWORKS UNDER SATURATION
CONDITION
129
should choose large L and w
to achieve a
reasonable V
and have acceptable delay and jitter
for a given network of size N. Otherwise the loss
rate will be higher, and V
will decline quickly.
3 CONCLUSIONS
In this paper, we discuss an inherent distortion of
video quality in IEEE 802.11 wireless networks.
Then we obtain a method to predict a quantitative
video quality requirement with proper setting of the
various design parameters of IEEE802.11 network.
Our approach is based on using a two-dimensional
Markov Chain models coupled with the use of ITU-
T Recommendation G.1070.
We also discuss how to optimize the parameters
setting of IEEE 802.11 network to minimize the
video distortion. For optimal parameters setting we
should choose large L and w
to achieve a certain
required V
for a given network size of N nodes.
REFERENCES
IEEE Std 802.11, 1999 Edition Part 11: Wireless LAN
Medium Access Control (MAC) and Physical Layer
(PHY) specifications.
ITU-T G.1070 "Opinion model for video-telephony
applications"
ITU-T J.241 "Quality of service ranking and measurement
methods for digital video services delivered over
broadband IP networks ".
Babich, F. D’Orlando, M. and Vatta, F., January 2008,
"Video Quality Estimation in Wireless IP Networks:
Algorithms and Applications" ACM Transactions on
Multimedia Computing, Communications, and
Applications (TOMCCAP) Volume 4, Issue 1.
Bianchi, G. March 2000, "Performance analysis of the
IEEE 802.11 distributed coordination function".
Selected Areas in Communications, IEEE Journal on
Volume 18, Issue 3.
Bouazizi, I. 2004, "Estimation of Packet Loss Effects on
Video Quality", First International Symposium on
Control, Communications and Signal Processing,.
Choi, L. Ivrlaˇc, M. Steinbach, E. and Nossek, J. 11-14
Sept. 2005, "Analysis of Distortion Due to Packet
Loss in Streaming Video Transmission over Wireless
Communication Links", IEEE International
Conference on Image Processing.
Khan, A., Sun, L. and Ifeachor, E. August 2009, "Content-
Based Video Quality Prediction for MPEG4 Video
Streaming over Wireless Networks", Journal of
Multimedia, Vol. 4, No. 4.
You, F. Zhang, W. and Xiao, J. 2009, "Packet Loss Pattern
and Parametric Video Quality Model for IPTV",
Eighth IEEE/ACIS International Conference on
Computer and Information Science.
Zhang, R., Regunathan, S. and Rose, K. Nov. 2001, "End-
to-end Distortion Estimation for RD-based Robust
Delivery of Pre-compressed Video" in Conf. Rec. 35th
Asilomar Conf. Signals, Systems and Computers,
Asilomar, CA.
Zhu, X. and Girod, B. Oct. 2008," Subjective Evaluation
of Multi-user rate Allocation for Streaming
Heterogeneous Video Contents over Wireless
Networks", in Proc. IEEE International Conference
on Image Processing (ICIP’08), San Diego, CA, USA
.
ICSOFT 2011 - 6th International Conference on Software and Data Technologies
130
