A Cross-layer Design for Video Transmission with TFRC in MANETs

George Adam

1,2

, Christos Bouras

1,2

, Apostolos Gkamas

Vaggelis Kapoulas

1,2

and Georgios Kioumourtzis

Computer Technology Institute & Press “Diophantus”, Patras, Greece

Computer Engineering & Informatics Dept, Univ. of Patras, Patras, Greece

University Ecclesiastical Academy of Vella, Ioannina, Greece

Center for Security Studies, P. Kanellopoulou 4, 10177, Athens, Greece

Keywords: MANETs, Multimedia, Video Transmission, TFRC, Cross-layer, AODV, SNR.

Abstract: Mobile Ad hoc NETworks (MANETs) are becoming more essential to wireless communications due to

growing popularity of mobile devices. However, MANETs do not seem to effectively support multimedia

applications and especially video transmission. In this work, we propose a cross-layer design that aims to

improve the performance of video transmission using TCP Friendly Rate Control (TFRC). Our design

provides priority to video packets and exploits information from the MAC layer in order to improve TFRC’s

performance. The proposed cross-layer mechanism utilizes Signal to Noise Ratio (SNR) measurements

along the routing path, in order to make the route reconstruction procedure more efficient. Simulation

results show that both the use of traffic categorization and the SNR utilization lead to important

improvements of video transmission over the mobile Ad hoc network. More specifically, simulations

indicate increased average Peak Signal to Noise Ratio (PSNR) for the received video, increased throughput

and packet delivery ration, as well as reduced average end-to-end delay.

1 INTRODUCTION

A fundamental difference in MANETs, compared to

other infrastructure-based wireless networks, is that

a mobile node could act also as a router while

having the possibility of being the sender or receiver

of information. The ability of MANETs to be self-

configured and form a mobile mesh network, by

using wireless links, make them very suitable for a

number of cases that other type of networks cannot

operate. An important usage scenario of MANETs

can be a disaster area or any kind of emergency, in

which the fixed infrastructure has been destroyed or

is very limited. However, this ability results in a

very dynamic topology in which routing becomes a

very complicated task. The impact of this dynamic

topology on multimedia applications and especially

on video streaming applications is high latency when

a wireless link breaks, and as a result routing

protocols should find alternate paths to serve

applications. Therefore, under these constrains there

should be in place additional mechanisms to

minimize latency in video streaming applications.

On the other hand, video streaming applications

use UDP as the transport protocol for video packets.

Although this is an obvious solution to avoid latency

caused by the retransmission and congestion control

mechanisms of TCP, it may cause two major

problems. The first one has to do with possible

bandwidth limitations in which uncontrolled video

transmission without any congestion or flow control

may lead to increased packet losses. The second

issue relates to TCP-friendliness. Under some

conditions uncontrolled video transmission may lead

to possible starvation of TCP-based applications

running in the same network.

The research community in order to address

these issues came with new proposals to provide

congestion control schemes based on those that are

already successfully implemented in TCP. However,

the proposed congestion control schemes are mainly

designed for use in wired networks, in which packet

losses primarily occur due to congested links. In

wireless networks the cause of packet losses is

mainly due to interference in the wireless medium.

Therefore, one needs to differentiate congestion

packets losses from random packet losses (Vazão et

al., 2008). To this direction a number of various

versions of TCP have been proposed including TCP

Veno (Cheng and Liew, 2003), TCP New Jersey

Adam G., Bouras C., Gkamas A., Kapoulas V. and Kioumourtzis G..

A Cross-layer Design for Video Transmission with TFRC in MANETs.

DOI: 10.5220/0004026200050012

In Proceedings of the International Conference on Data Communication Networking, e-Business and Optical Communication Systems (DCNET-2012),

pages 5-12

ISBN: 978-989-8565-23-5

 2012 SCITEPRESS (Science and Technology Publications, Lda.)

(Xu, Tian and Ansari, 2005) and TCP NCE

(Sreekumari and Chung, 2011). In another work

(Shagufta, 2009), the impact of TCP variants on the

performance in MANETs routing protocols is

investigated.

The most well-known congestion control

mechanism that can be used on top of other transport

protocols, such as UDP, is the TCP-friendly

Congestion Control (TFRC) (Handley, Floyd,

Padhye and Widmer, 2008), which is already an

international standard. However, even TFRC is

facing some limitations in wireless environments

and especially in MANETs. In (Chen and Nahrstedt,

2004) these limitations are studied and it is shown

that TFRC can be used in MANETs only when strict

throughput fairness is not a major concern.

Moreover, they analyze several factors contributing

to TFRC’s conservative behaviour, many of which

are inherent to the MANET network. While their

study reveals the limitations of applying TFRC to

MANETs, they address the open problem of

multimedia streaming in these networks and propose

an alternative scheme based on router’s explicit rate

signalling and application’s adaptation policies.

In order to overcome the above limitations an

algorithm is proposed in (Li, Lee, Agu, Claypool

and Kinicki, 2004), which is termed as Rate

Estimation (RE) TFRC, and it is designed to

enhance TFRC performance in wireless Ad hoc

networks.

A large variety of research has been conducted

regarding the usefulness of the wireless medium-

related metrics. In (Zhang et al., 2008) a

systematically measurement-based study on the

capability of SNR is performed to characterize the

channel quality. Although it is confirmed that SNR

is a good prediction tool for channel quality, there

are also several practical challenges.

Our motivation in this paper is to address the

aforementioned technical issues, making video

streaming in MANETs a promising application area.

To this direction, we design a cross-layer

mechanism that:

 Provides priority to video packets against other

data packets.

 Implements TFRC to provide congestion and

transmission rate control to video applications.

 Enhances routing operation with additional

wireless medium-related metrics in order to

improve the wireless transmission performance.

The proposed cross-layer mechanism is tested

through simulations.

The rest of the paper is organized as follows: In

section 2 we discuss the cross-layer design. In

section 3 we provide an analysis of the proposed

mechanisms. In Section 4 we briefly discuss the

simulation environment under which we evaluate

our cross-layer design. In Section 5 we present the

simulation results. Finally, we conclude the paper in

Section 6 with plans for future work.

2 CROSS-LAYER DESIGN

The proposed cross-layer design is based on the

attributes of voice and video streaming applications,

which are characterized by different tolerance in

terms of end-to-end delay. A real time service, like

video transmission, requires much less delay jitter

values than a file transfer application. A way to

minimize delay jitter is to prioritize traffic and to

adapt the routing procedures depending on

application requirements. The proposed cross-layer

design invokes three layers in which we apply our

adaptations

At the MAC layer, we differentiate the access of

various applications with the use of the IEEE

802.11e protocol (IEEE Std., 2005), based on

Quality of Service (QoS) criteria. Therefore, the IP

packets are marked based on the underlying

application type. This is a simpler task in mesh

networks than in wired networks with fixed

infrastructure, in which different administrative

domains may exist in a path between video sender

and receiver(s). Ad hoc networks provide this

flexibility as every node in the network acts also as

router. The main function for providing QoS support

in IEEE 802.11e protocol is the Enhanced

Distributed Coordination Function (EDCF). This

function is responsible for managing the wireless

medium in the Contention Period (CP) and enhances

the Distributed Coordination Function (DCF)

function of the legacy IEEE 802.11 protocol. The

priority of each Traffic Class (TC) is defined by the

following parameters:

 The transmission opportunity (TXOP), which

stands for “the time interval when a station has

the right to initiate transmission, defined by a

starting time and the maximum duration”. It is

measured in milliseconds.

 The Arbitration Interframe Space (AIFS), which

is at least DCF Interframe Space (DIFS) long.

When the AIFS is represented by a number n

instead of time, it is calculated according to the

following equation:

IFS SIFS n SlotTime=+

(1)

DCNET2012-InternationalConferenceonDataCommunicationNetworking

 The minimum value of the Contention Window

(CW)

 The Persistence Factor (PC), which is used to

increase the CW after any unsuccessful

retransmission and this CW is different for each

TC.

The source code (Wiethölter and Hoene, 2003)

used in this work is compliant with the

specifications of the IEEE 802.11e protocol but

supports only up to four different data Traffic

Categories (TCs). In the latest IEEE 802.11e

standard, the protocol can support up to eight

different TCs but we regard the current

implementation with four TCs for our work as

sufficient enough. Table 1 outlines the different QoS

parameters for the four TCs.

Table 1: QoS parameters for the four TCs in IEEE

802.11e.

TC[0] TC[1] TC[2] TC[3]

PF 2 2 2 2

AIFS 2 2 3 7

CW_MIN 7 15 31 31

CW_MAX 15 31 1023 1023

TXOP limit 0.003 0.006 0 0

At the network (routing) layer we utilize SNR

information for improving the routing performance.

We use the Ad hoc On-Demand Distance Vector

(AODV) (Perkins and Belding-Royer, 2003) routing

protocol which is among the most popular ad hoc

routing protocols and is capable for both unicast and

multicast routing. AODV is a reactive routing

protocol that is based on the Bellman-Form

algorithm. In general, the reactive protocols search

for a routing path between nodes only on demand.

The advantage of this method is that utilizes low

network bandwidth and does not introduce routing

overhead when data transmission is not required. In

contrast, proactive protocols establish and maintain

routing paths for nodes even if there is no need to

transfer data. This allows lower latency but requires

higher network management cost.

AODV uses originator and destination sequence

numbers to avoid both “loops” and the “count to

infinity” problems that may occur during the routing

calculation process. As a reactive routing protocol, it

does not explicitly maintain a route for any possible

destination in the network. However, its routing

table maintains routing information for any route

that has been recently used, so a node is able to send

data packets to any destination that exists in its

routing table without flooding the network with new

Route Request messages. In cases where the

mobility is high, the routing paths need to be

reconstructed frequently. For this purpose, we

introduce a mechanism that utilizes the SNR

measurements along the routing path, in order to

make the reconstruction procedure more efficient. In

essence, the reduction of the measured SNR, may

signify that the relative nodes are travelling further

apart from each other and a disconnection of the link

between them is eminent. At this stage the cross-

layer design enables in advance the route

reconstruction process to avoid the temporary

disconnection.

At the application (APP) layer we implement

TFRC for congestion control with enhanced

functions to improve the estimations of TFRC and to

better utilize the available bandwidth. To do so, we

use feedback information from the receiver. The

TFRC feedback packet is modified in order to

include the SNR measurements along the routing

path. Moreover, we consider rate adaptive video

transmission for scaling among different qualities to

achieve better bandwidth utilization. This adaptation

is also achieved by utilizing the reception rate and

packet loss estimation that TFRC feedback provides.

The proposed cross-layer design with adaptations

at the MAC, Network and Application layers is

depicted in Fig. 1.

Figure 1: Proposed cross-layer design.

3 MECHANISM ANALYSIS

TFRC is a congestion control mechanism which is

designed for unicast flows that compete with TCP

traffic. Compared to TCP, TFRC has lower variation

of throughput over time, so in many cases is more

ACross-layerDesignforVideoTransmissionwithTFRCinMANETs

suitable for multimedia applications. However,

TFRC should be used when there is a need for

smooth throughput as it responds slower than TCP

to changes in the network conditions. It is designed

for rate adaptive applications that use fixed size

packets and can increase or decrease the sending

rate. TFRC is a receiver-based mechanism which

means that the congestion control information is

calculated at the receiver side and then it is sent to

the sender using a feedback message. Our proposed

adaptation extends this feedback message with SNR

information.

The use of the SNR measurements in MANETs

is not straightforward. A transmission path in a

multi-hop topology consists of many single links

with different quality. This heterogeneity is affected

by nodes' hardware or the distance of each one

wireless link. Therefore, one difficulty is that there

are more than one SNR measurements that can be

exploited, but there is no provision in the existing

protocols to “carry” this information along with

other information to the sending and receiving

nodes. However, once a routing path is established

then the transmission quality can be degraded even

if only a single link of the multi-hop communication

is degraded. In this environment, the link with the

lowest quality directly affects the total quality of the

routing path. To overcome the above difficulty, the

proposed mechanism maintains only the minimum

SNR measurement along the multi-hop path (which

can be more easily attached to a packet with video

information). This information is then made

available to TFRC protocol and it is included in the

next feedback report, so that both sender and

receiver are aware of the link quality. The feedback

message contains the following information:

 The timestamp of the last data packet received.

 The delay between the last received data packet

and the generation of the feedback report.

 The rate at which the receiver estimates that

data was received since the last sent feedback

report.

 The receiver’s current estimate of the loss event

rate.

 Minimum SNR along the routing path.

The TFRC feedback report is utilized to adapt

the rate of the video transmission and also to

maintain the routing path quality to high levels. For

this purpose, the proposed mechanism implements a

TFRC feedback handling algorithm (Algorithm. 1).

Firstly, the mechanism extracts the receiver address

and the minimum found SNR and then a comparison

with a predefined SNR threshold is made. If the

received SNR is found to be lower than the

threshold, meaning that the end to end connection is

likely to be lost, then a new route discovery

procedure is initiated. Moreover, a simple timer is

exploited in order to avoid very frequent routing

path discoveries. This means that a new discovery

procedure is allowed to be executed only if the timer

has expired.

Algorithm 1: Modified TFRC feedback handling

algorithm.

ModifiedRecvTfrcFeedback(feedback_packet) {

snr = get_snr(feedback_packet);

receiver_address = get_source_address(feedback_packet);

if (snr < SNR_THRESHOLD and

TIMER_EXPIRED = true) {

routing_record = routing_table.lookup(receiver_address);

schedule_update_routing_path(routing_record);

}

X_recv = data_reception_rate(feedback_packet);

p = estimated_loss(feedback_packet);

adapt _transmission(receiver_address, X_recv, p);

}

4 SIMULATION ENVIRONMENT

For the simulation experiments the ns-2 simulator is

used. The simulation environment is extended in

order to support the mechanisms which described in

the previous section.

In order to conduct a number of realistic

experiments with real video files we use the

Evalvid-RA (Lie and Klaue, 2008) tool-set in

conjunction with ns-2. Evalvid-RA is a framework

and tool-set to enable simulation of rate adaptive

VBR video. It has the capability to generate true rate

adaptive MPEG-4 video traffic with variable bit rate.

The tool-set includes an online (at simulation time)

rate controller that, based on network congestion

signals, chooses video quality and bit rates from

corresponding pre-processed trace files.

Figure 2: Simulation-time rate controller of Evalvid-RA.

DCNET2012-InternationalConferenceonDataCommunicationNetworking

As shown in Figure 2, the Evalvid-RA rate

controller that is executed at simulation time chooses

correct frame sizes (emphasized boxes) from

different trace files. These files represent different

video qualities for each quantizer scale. The same

figure shows an example of a video transmission

with 25 fps and three pre-processed qualities. The

GOP size is 2 with the sequence of one I and one P

frame.

For our simulations, we use a YUV raw video,

which consists of 9144 frames and has duration of

366 seconds. We encode this raw video with the

ffmpeg (Tomar, 2006) video encoder to produce an

MPEG-4 video file. The frame size is set to 176x144

pixels, which is known as the Quarter Common

Intermediate Format (QCIF). The temporal

resolution is set to 25 frames per second with Group

of Pictures (GoP) size equal to 12. After the

simulation, we reconstruct the received video file

and perform a frame-by-frame comparison between

the original transmitted and the received video file in

order to evaluate the quality of the received video.

The mobility model that is studied is based on

the Manhattan city model with uniform sized

building blocks. Manhattan grid mobility model can

be considered as an ideal model to represent the

topology of a big city. The simulation area is

500x500 meters in a 5x5 grid. Inside this area, there

are 50 mobile nodes representing moving vehicles

that are actually the transmitters and receivers of the

information. The moving speed varies from 0 to

10m/sec, having a mean value of 4m/sec.

The simulations include some low rate

background traffic between the moving nodes. Each

node transmits in Constant Bit Rate (CBR) mode an

amount of 2,560 bytes per second. Table 2

summarizes the simulation parameters that are used.

Table 2: Simulation parameters.

Mobility model Manhattan Grid Model

Simulation duration 366 seconds

Number of nodes 50

Simulation area 500 x 500m

Node speed 0 – 10 m/sec (random)

Antenna OmniAntenna

Data rate 2Mbps

Video bitrate 32kbps – 2Mbps (variable)

5 PERFORMANCE EVALUATION

The performance of the proposed cross-layer

mechanism is evaluated in three scenarios.

 First, we evaluate the video transmission

without any traffic prioritization in the MAC

layer.

 Then, we introduce the IEEE 802.11e protocol

in order to prioritize the video traffic against the

background traffic.

 The last simulation utilizes the SNR mechanism

for further performance enhancement.

A number of simulations have been conducted,

in order to investigate the affect of the SNR

threshold on the perceived video quality by the end

user. For this purpose we calculate the Peak Signal

to Noise Ratio (PSNR) by directly comparing the

video file sent by the sender with the same file at the

end user on a frame-by-frame basis. Equation 2

gives the definition of PSNR between the luminance

component Y of source image S and the destination

image D:

()()

() 20log

,, ,,

where

2 1, number of bits per pixel (luminance component)

col row

peak

col row

peak

PSNR n

Y nij Y nij

⎛⎞

⎜⎟

−⎡⎤

⎣⎦

⎜⎟

⎝⎠

=− =

∑∑

(2)

The selection of SNR threshold affects the

efficiency of the routing path reconstruction.

Choosing a low threshold may result to very late

reconstruction, while choosing a high threshold may

result to very frequent route discovery processes that

will add routing overhead to the ad-hoc network.

For the evaluation of the performance of the

proposed cross-layer design, we examine the PSNR

of the received video, with respect to the original

video, the average throughput, the packet delivery

ratio, and the average end-to-end delay. The

simulation results show that both the use of traffic

categorization and the utilization of SNR mechanism

lead to important improvements, in all the above

metrics, during the video transmission over the

MANET network.

More specifically, Figure 3 shows the PSNR

measurements among different SNR thresholds. For

the rest of the simulation, the SNR threshold is

chosen to be 33.0 dB. It should be mentioned that

the above PSNR measurements suggest a SNR

threshold which may not be suitable for all network

topologies and network conditions. The above PSNR

measurements must be used in order to estimate the

SNR threshold when someone plans to used the

ACross-layerDesignforVideoTransmissionwithTFRCinMANETs

proposed mechanism in a network.

Average PSNR

3232.53333.53434.53535.5

SNR Thre shold

PSNR (db)

Figure 3: Average PSNR among different SNR thresholds.

As the cross-layer design intends to improve

video transmission, the performance evaluation is

focused in video related metrics.

In figure 4 the average PSNR is displayed for the

three simulated scenarios. We can observe that the

use of traffic categorization (with the use of

802.11e) leads to a small improvement of average

PSNR but the utilization of the SNR mechanism

leads to a significant improvement (more than 1.5

dB comparing with 802.11g) which is an important

result.

Average PSNR

37.5

38.5

39.5

802.11g 802.11e 802.11e + SNR

utilization

PSNR (db)

Figure 4: Average PSNR.

This means that for all kinds of frames the

implementation of the proposed cross-layer design,

that includes SNR measurements to better estimate

the link quality, greatly reduces the video frame

losses, and thus allows for a better video

reconstruction at the receiver side. It is worth noting

that without the implementation of the cross-layer

design, the frame losses are at a level in which video

reconstruction may not be possible at all at the

receiver side. In contrast, the frame losses when the

proposed cross-layer design is implemented are at

level where video reconstruction can be done with

only a few disruptions.

Figure 5 shows the average throughput during

the three evaluation scenarios. Again the use of

traffic categorization (with the use of 802.11e) leads

to an improvement of throughput and the utilization

of the SNR mechanism further leads to a significant

additional improvement of throughput (more than

100Kbps comparing with 802.11g). We have to

mention that the improvement in throughput is

significant in terms of QoS from the end user

perspective (PSNR measurements in Figure. 4)

because a small increase in throughput can lead to

significant improvement of the end user experience.

Throughput

840

860

880

900

920

940

960

980

1000

802.11g 802.11e 802.11e + SNR

utilization

Throughput (kb/s)

Figure 5: Average throughput.

Figure 6 shows the packet delivery ratio during

the three evaluation scenarios. Similar conclusions

as in the case of the average throughput can be

inferred. Again, the use of traffic categorization

(with the use of 802.11e) leads to a significant

improvement of packet delivery ratio and the

utilization of SNR mechanism leads to a small

additional significant improvement of the packet

delivery ration.

Packet Delivery Ratio

98.2

98.4

98.6

98.8

99.2

99.4

802.11g 802.11e 802.11e + SNR

utilization

Packet Delivery Ratio (%)

Figure 6: Packet delivery ratio.

DCNET2012-InternationalConferenceonDataCommunicationNetworking

Finally, figure 7 shows the average end-to-end

delay during the three evaluation scenarios. Both the

use of traffic categorization (with the use of

802.11e) and the utilization of the SNR mechanism

lead to a significant improvement of average end-to-

end delay. We have to mention that the above

improvement in average end-to-end delay is very

important for video streaming applications.

Average end-to-end delay

802.11g 802.11e 802.11e + SNR

utilization

Average end-to-end delay (ms)

Figure 7: Average end-to-end delay.

The above results show that the use of both the

traffic categorization and the SNR-utilizing cross-

layer mechanism lead to important improvements

during the transmission of multimedia data over the

MANET network. This improvement can lead to a

noticeable quality improvement of the video

received, as subjectively judged by some viewers.

This judgment verifies that the improvement can

also be perceived by the users.

In addition the above results indicate that the use

of the cross-layer design, can lead to significant

improvements in video transmission in MANETs.

These improvements can help make the difference in

MANETs between an interrupted, low-quality video

transmission and a usable video transmission service

without perceived annoyances for the users..

6 CONCLUSION AND FUTURE

WORK

We presented in this work a cross-layer design that

aimed to improve the performance of video

transmission with the use of TFRC. Our design

provided priority to video packets and exploited

information from the MAC layer (SNR) in order to

improve the TFRC performance. Simulation results

showed that the proposed cross-layer design led to

improving performance, under several metrics, and

could result in perceived improvements of the

received video quality.

We also showed how a cross-layer design involving

the Application, Network and the MAC layers can

improve QoS in MANETs by sharing information

between non-adjacent layers.

Our future work includes the implementation of

an adaptive estimation for the appropriate SNR

threshold based on network metrics. In addition, we

plan to implement a prototype of the proposed cross

layer design and evaluate it in a real MANET.

Another interesting extension of this work is to

use the SNR measurements in order to provide

TFRC with estimations of whether the observed

packet loss is due to network disruption or due to

congestion. This is expected to have a positive

impact on the performance of TFRC and the video

transmission in MANETs, as well.

Furthermore, we plan to investigate the

(combined) use of new cross-layer designs and

mechanisms in order to come up with a balanced set

of improvements that could provide the best

outcome. Finally, we plan to investigate the effect of

the proposed design, and especially the use of SNR,

in the performance of other routing protocols in

MANETs.

REFERENCES

Vazão, T., Freire, M., Chong, I., Yan, J., Zheng, X., Liu,

J., Li, J., 2008. “Wireless Loss Detection for TCP

Friendly Rate Control Algorithm in Wireless

Networks”, Information Networking. Towards

Ubiquitous Networking and Services, Lecture Notes in

Computer Science, Springer Berlin / Heidelberg, 113-

122

Cheng, P. Fu, L., S. C., 2003. "TCP Veno: TCP

enhancement for transmission over wireless access

networks," Selected Areas in Communications, IEEE

Journal on, vol.21, no.2, pp. 216- 228.

Xu, K., Tian, Y., Ansari, N., 2005. “Improving TCP

performance in integrated wireless communications

networks”, Computer Networks, Volume 47, Issue 2,

Wireless Internet, ISSN 1389-1286, DOI: 10.1016/

j.comnet.2004.07.006., Pages 219-237.

Sreekumari, P., Chung, S., 2011. “TCP NCE: A unified

solution for non-congestion events to improve the

performance of TCP over wireless networks”,

EURASIP Journal on Wireless Communications and

Networking 2011, Published: 29 June 2011.

Shagufta H., 2009. “A Throughput Analysis of TCP

Variants in Mobile Wireless Networks”, In

Proceedings of the 2009 Third International Confe-

rence on Next Generation Mobile Applications, Servi-

ces and Technologies (NGMAST '09). IEEE Computer

Society, Washington, DC, USA, 279-284.

ACross-layerDesignforVideoTransmissionwithTFRCinMANETs

Handley, M., Floyd, S., Padhye, J., Widmer, J., 2008. RFC

3458, “TCP Friendly Rate Control (TFRC): Protocol

Specification”.

Chen, K., Nahrstedt, K., 2004. “Limitations of Equation-

Based Congestion Control in Mobile Ad Hoc

Networks”. In Proceedings of the 24th International

Conference on Distributed Computing Systems

Workshops - W7: EC (ICDCSW'04) - Volume 7

(ICDCSW '04), Vol. 7. IEEE Computer Society,

Washington, DC, USA, 756-761.

Li, M., Lee, S. C., Agu, E., 2004. Claypool, M. Kinicki,

R., “Performance Enhancement of TFRC in Wireless

Ad Hoc Networks”, In Proceedings of the 10th Inter-

national Conference on Distributed Multimedia

Systems (DMS), Hotel Sofitel, San Francisco, Califor-

nia, USA, September 8 - 10, 2004.

Zhang, J., Tan, K., Zhao, J., Wu, H., Zhang, Y., 2008. "A

Practical SNR-Guided Rate Adaptation", INFOCOM

2008. The 27th Conference on Computer Communi-

cations. IEEE, vol., no., pp. 2083-2091, 13-18 April

2008.

IEEE Standard for Information technology, 802.11e-2005,

Local and metropolitan area networks-- Specific

requirements-- Part 11: Wireless LAN Medium Access

Control (MAC) and Physical Layer (PHY) Specifica-

tions - Amendment: Medium Access Method (MAC)

Quality of Service Enhancements, 2005.

Wiethölter, S., Hoene, C., 2003. "Design and Verification

of an IEEE 802.11e EDCF Simulation Model in ns-

2.26", Technical Report TKN-03-019, Telecommunica-

tion Networks Group, Technische Universität Berlin.

Perkins, C., Belding-Royer, E., 2003. Ad hoc On-Demand

Distance Vector (AODV) Routing, RFC 3561.

Lie, A., Klaue, J., 2008. “Evalvid-RA: Trace Driven

Simulation of Rate Adaptive MPEG-4 VBR Video”,

Multimedia Systems, Springer Berlin / Heidelberg,

Volume 14, Number 1, pp. 33-50.

Tomar, S., 2006. Converting video formats with FFmpeg,

Linux Journal, v.2006 n.146, p.10.

DCNET2012-InternationalConferenceonDataCommunicationNetworking