Documente Academic
Documente Profesional
Documente Cultură
Abstract
In this article, video traffic H.264/SVC and evaluation QoS metrics analyzed by Delay
and Throughput in relation to the number of jumps performed by the AODV protocol in
an ideal environment (without traffic) and no Ideal with DCF and EDCA traffic.
Particularly QoE obtain advantages from the point of view of the user, by encoding,
transmitting and decoding video in a simulation environment as NS2, by evaluating the
video framework myEvalSVC illustrated. In this way we can analyze the impact of the
transmission and retrieval of video in uncontrolled environments, such as Ad Hoc
networks with QoS and without QoS.
Key Words: H264/SVC, QoS, QoE, DCF, EDCA
1. INTRODUCTION
With the increase of video streaming on demand and in real time, with the proliferation of
video services over the Internet and with the exponential growth of mobile devices capable
of processing multimedia content, the wireless video communication is become an attractive
market niche, which is receiving attention from industry and academy as applications of
wireless video transmission every day are easier to implement and integrate multiple devices
with Wi-Fi.
Applications in Wireless Local Area Networks (WLAN), support video
streaming technologies (VoIP, IPTV, etc.). Their study is attractive due to mobility and
portability that wireless ad hoc networks offer as an alternative to infrastructure
16110
networks, since the delivery of real time video imposes strict requirements on time and
bandwidth, emerging many problems as a Quality of Service (QoS) means.
This article describes how the basic medium access mechanism Distributed
Coordination Function (DCF) and Enhanced Distributed Channel Access (EDCA),
H.264/SVC and scalability analysis of jumps that presents the protocol Ad hoc
On-Demand Distance Vector (AODV), in order to obtain a quantitative and qualitative
analysis of the behavior of the video in Ad hoc networks, through simulations with the
framework built into NS2 myEvalSVC.
The myEvalSVC [1] framework to estimate the quality of video transmissions
H.264/SVC in NS2, supported on three main processes: encoding, transmission and
decoding video. For analysis of the observations QoS time constraints presented
metrics Throughput Delay and used. As for Quality of Experience (QoE) is concerned,
the visual comparison of the transmitted video regarding the received video and
quantification of Video Quality Metrics through the Peak Signal-to-Noise Ratio
(PSNR) is analyzed.
16111
channel is busy performing a request repeatedly until a node finds that the channel is free and
as a result of the above process collisions occur. DCF, in order to avoid such collisions
specifies a random Backoff, forcing a node to defer its access to the channel for additional
time interval determined by the following expression [4]:
Backoff Tiempo = Random (0,CW) x Slot_Time
where CW is contention window.
(1)
Each access category has its own transmit queue characterized by the parameters
of network traffic. The prioritization between different categories of access is obtained
by suitably configuring traffic parameters in each queue. A scheme operating system
access categories shown in Figure 1, the parameters of interest are:
Arbitrary Inter-Frame Space Number (AIFSN): It is the minimum time interval
since the physical medium is detected as empty or available until the transmission
begins [6].
AIFS [ AC ] SIFS AIFSN [ AC ] slot _ time
(2)
16112
At the time that the data packets arrive at the MAC Service Access Point (SAP),
the MAC layer handles 802.11e properly classify incoming traffic, sending them to the
appropriate queue of the MAC Service Data Unit (MSDU) layer. Each of these units of
the different queues AC (Access Category) competes internally by the TXOP.
The algorithm internal struggle calculates the backoff timeout independently for
each AC line, according AIFSN, CW parameters and a random number. The timeout
mechanism is similar to DCF and tail with the lowest backoff win in the internal
competition.
The AC queue winner externally compete for access to the wireless medium. The
algorithm external strife no significant changes compared with DCF, DCF except that
the Backoff and timeouts are fixed, in contrast to 802.11e, which are variable and are
properly configured according to the corresponding AC queue.
Through proper setting of the parameters of the AC queue, traffic performance is
tight and can be achieved by prioritizing traffic. This requires a QoS Capable Access
Points (QAP) to maintain a common set of parameters in the queues and guarantee fair
access between the different stations that compose the QoS Enabled Basic Service Set
(QBSS) network. Similarly, in order to adjust the asymmetry between upstream
traffic/incoming qsta to the QAP and down/outgoing to QSTA QAP, a separate set of
EDCA parameters defined exclusively for the QAP [7]. Figure 2 compared the medium
access mechanism EDCA referring to DCF.
16113
16114
Overall SVC can provide three types of scalability: temporal, spatial and SNR
quality, allowing multiple video representations thus adapting to the speed and quality
levels for streaming video.
16115
In SVC, the bitstream is divided into a base layer and one or more enhancement
layers. The base layer is considered more important than the enhancement layers, as in
this layer needs less transmission bandwidth and the minimum acceptable contains
video information. Enhancement layers require more bandwidth for the definition and
enhancement of details that enhance its quality. Figure 3 is a schematic representation
of scalabilities SVC and Figure 4 shows the H.264/SVC Layered Structure.
Spatial scalability refers to the ability to represent the same video in different
resolutions and screen sizes, including: QCIF, CIF and 4CIF. In general, the spatially
encoded scalable video using images in the space used by the lower layers as a
prediction of the higher layers in order to further improve coding efficiency.
Temporal scalability refers to the possibility of representing the same video at
different temporal resolutions and frame rates, ie, the number of frames contained in the
first second of the video allow the video to be played at different frame rates. Usually
done using temporary sample images from a lower layer as a top layer prediction.
Quality scalability, also called scalability of signal to noise ratio (SNR), refers to
the possibility of representing the same video at different levels of perceptual quality.
The SNR scalable encoding coefficients quantized Discrete Cosine Transform (DCT) at
different levels of accuracy by using different quantization parameters.
3. myEvalSVC
myEvalSVC is a framework for evaluating H.264/SVC built into NS2, based on
Scalable Video-streaming Evaluation Framework (SVEF) [10], which allows evaluating
network topologies, architectures and routing protocols realistically. myEvalSVC
process starts by encoding YUV video format, in which you can set encoding
parameters (temporal, spatial, or combined SNR). Once selected encoding parameters,
we proceed to pass the video modules BitStreamExtractor which is developed by JSVM
and developped by FN Stamp SVEF, in order to generate a trace file NALU [11].
With the trace file NALU is necessary to create a compatible file in NS2, for this a
filter developed by myEvalSVC prepare_sendtrace.awk, which delivers a trace is used
completely compatible and integrated in the simulator. For reading and use of this
traceis necessary declare a myEvalSVC agent, which allows to design and evaluate the
performance of the SVC transmission topologies, protocols, and network architectures.
Within the simulator must be configured myEvalSVC_Sink their Agent, which is
used to receive packets SVC that record the time, the packet size, number of frame in a
trace file output. With the information in this trace file can be extracted by creating
filters awk calculate different network metrics. At this point can be evaluated QoS and
performance of any scenario that video transport.
To evaluate QoE from the point of view of the user, myEvalSVC decode the video
that came on the receiver or target node, it is necessary to begin the process of decoding.
The decoding process starts by filtering the trace file using the filter
prepare_receivedtrace1.awk and prepare_receivedtrace2.exe, both developed by
myEvalSVC. The resulting file is processed by the SVEF module, which generates a
trace file that deletes NALU packets that arrived too late and could not be decoded.
After filtering the NALU trace output file is sent to BitStreamExtractor module, which
16116
generates a trace of H.264, which is then filtered by the JSVM decoder that generates a
file YUV video output. At this point QoE can be evaluated through the comparison of
input PSNR and output PSNR and by reconstructing the video [1].
4. PERFORMANCE METRICS
Inside the transmission of streaming video are the most important constraints related to
time and bandwidth, therefore, the Delay and Throughput is highlighted as measures
overall performance, which analyze the behavior of video within Ad Hoc networks. The
metric to be analyzed are:
4.1. Average end-to-end delay
The Average end-to-end delay [12] is defined as the time difference between the instant
when a packet is received by the destination node and the instant of time that has been
sent by a source node. Within the video transmission applications must comply with the
standard QoS and packet delay must be limited and decreased for high performance
transmission. The Average end-to-end delay has been calculated as shown in Equation
3:
H
TAVG
i
r
H ti
i 1
16117
(3)
Nr
LC
R f ( )
L
(4)
LC
R f is the payload transmission rate, R is the Binary transmission rate
L
in bits/seconds and f ( ) is the packet success rate defined as the probability of
receiving a packet correctly. This probability is a function of the signal-to-noise ratio
( ) .
where,
16118
10us
20us
50us
32
1024
192bits
224bits
112bits
1Mbps
1Mbps
0.2Mbps
0.3Mbps
5s
1000m x 1000m
250m
50
AODV
6m/s
The test sequence used corresponds to video file "Foreman" [15] YUV CIF
(352X288) with 300 frames and encoded by JSVM where temporal scalability to the
results presented in this article is enable.
16119
One of the main features of myEvalSVC lies video recovery in the target node,
appreciating the initial delay that causes the network. In case of packet loss
myEvalSVC replaces the lost image of the last image who arrived correct package.
The Peak Signal-to-Noise Ratio (PSNR) may be considered as a QoE metric,
since the quality of the received image is evaluated. In a video stream H.264/SVC base
layer is transmitted and sometimes finds that packet loss can be lost enhancement
layers. Although to the naked eye many of these losses are not significant, the PSNR
provides a quantitative measure of the quality of video that is being perceived. This
paper provides the results for a jump is shown by the high number of simulations.
16120
Fig. 8: PSNR original video with DCF and EDCA traffic for a traffic jump
Background.
The routing protocol used for the assessment is Ad hoc On-Demand Distance
Vector (AODV), used by all mobile nodes in the proposed network, which offers the
following features [14]:
16121
The above results show a proportional increase between the number of hops and
delay for a constant speed of 6m/s, finding that 90% of the video is encoded for the same
scenarios or less than two jumps.
For scenarios equal to or greater than the three top video decoding than 10% and
therefore higher packet loss, which is reflected in the persepcion QoE is presented.
In the figure 10 shown throughput metrics and evaluation of network performance
at a constant speed and end to end delay is analyzed, noting that traffic packets to
16122
transmit video H.264SVC a DCF mechanism in his first two jumps delay is low and
adapts its throughput to Best Effort traffic because all packets in DCF are treated with
the same priority.
By contrast, when the video traffic is transmitted in a H.264SVC EDCA
mechanism allows prioritize packets Best Effort traffic and traffic with a lower priority
Background, giving precedence first to the voice traffic and secondly to traffic video.
Therefore, EDCA may reach rates of less than 10ms delay obtaining a bandwidth of
0.6Mbps and a jump two 0.26Mbps obtained jumps where video packet loss H.264SVC
less than 10% (with traffic Best Effort and Background) network ensure proper
decoding at the target node.
However, the ideal traffic (no Background and Best Effort traffic) in DCF is
comparable to the results obtained with EDCA traffic.
Figure 8 shows PSNR values for different access methods. The upper curve
represents the ideal values of PSNR NALU lossless (original video), the other two
curves correspond to the transmission of video using EDCA and DCF, noting that DCF
has less video Quality of EDCA, this interpretation traduciendoce QoE for user.
8. CONCLUSIONS
In this paper, QoS and QoE evaluated in providing video H.264SVC in WLANs through
the DCF and EDCA mechanisms, assessing the quality of streaming video, five primary
scenarios were simulated by highlighting the number of jumps performed by the AODV
protocol to deliver video to a target node. The results showed that reducing the
throughput depending on the number of hops decreases the available bandwidth while
maintaining the video until the third and fourth jump with only EDCA mechanism,
except that the DCF mechanism with the same throughput is maintained until the third
and fourth jump, but lost 10% superiosres video, making the video encoding is not
complete with myEvalSVC Framework.
These results show that the transmission H.264/SVC on EDCA can achieve better
PSNR values of the transmission through DCF, illustrating how the throughput and
delay as a function of the number of hops can be percevida by the end user, through the
effects visual myEvalSVC the Framework allows for the recostruir video lens or
receiving node. This feature is one that highlights the myEvalSVC Framework as a tool
for evaluation of QoE, for an end user, which can visually compare the performance of
the video in a different network topologies.
Increasing delay and throughput decrease with increasing number of hops,
concludes that the video can be decoded H.264SVC rates 10% lower losses in the case
of DCF, EDCA provides QoS while Ad Hoc Network 3 and 4 hops for the particular case
of 50 nodes that maintain a constant speed of 6m / s and with 8 nodes act as traffic
sources with their respective target nodes/receiver.
16123
9. BIBLIOGRAPHY
[1]
16124
[13] Li, J., Blake, C., De Couto, D. S., Lee, H. I., & Morris, R. (2001, July). Capacity of
ad hoc wireless networks. In Proceedings of the 7th annual international
conference on Mobile computing and networking (pp. 61-69). ACM.
[14] Klein-Berndt, L. (2001). A quick guide to AODV routing. Wireless
Communications Technologies Group, NIST (http://w3. antd. nist.
gov/wctg/aodv_kernel/).
[15] Video Traces Research Group. (2007). YUV 4: 2: 0 Video Sequences. Arizona
State University, Available online at: http://trace. eas. asu. edu/yuv/index. html.
Paulo Mendes.