Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
Hindawi Publishing CorporationInternational Journal of Digital Multimedia BroadcastingVolume 2013 Article ID 474212 7 pageshttpdxdoiorg1011552013474212
Research ArticleBandwidth Management in Wireless Home Networks forIPTV Solutions
Tamaacutes Jursonovics1 and Saacutendor Imre2
1 Deutsche Telekom AGmdashProducts amp Innovation T-Online-Allee 1 64295 Darmstadt Germany2Department of Telecommunications Budapest University of Technology Budapest Hungary
Correspondence should be addressed to Tamas Jursonovics jursonovicstgmailcom
Received 29 October 2012 Accepted 8 January 2013
Academic Editor Massimiliano Laddomada
Copyright copy 2013 T Jursonovics and S Imre This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited
The optimal allocation of the retransmission bandwidth is essential for IPTV service providers to ensure maximal servicequality This paper highlights the relevance of the wireless transport in todayrsquos IPTV solution and discusses how this new mediaaffects the existing broadcast technologies A new Markovian channel model is developed to address the optimization issuesof the retransmission throughput and a new method is presented which is evaluated by empirical measurements followed bymathematical analysis
1 Preface
The terminology of Open IPTV Forum (OIPF) specificationrelease 2 [1] has been chosen to describe the internet protocoltelevision IPTV features in this paper because we experiencea wide diversity of terms in several journals which mayconfuse the reader We believe that the OIPF terms arestraightforward and they can be easily interpreted on anyIPTV solutions although our work is independent from thestandard itself
2 Motivation
The telecommunication industry is tardily changing Theemerging market of the new generation over-the-top (OTT)services from Google Microsoft Apple or Amazon had puta big pressure on operators tomove away from the traditionaltelecom model and assess threats and opportunities fromOTT players There is a big race for the customers todayand legacy industry has to extend its portfolio with variousvalue adding services like triple-play rich communicationor mobile payment This paper evaluates one specific topic ofthis competition the video broadcasting services
We have observed the rapid evolution of the IPTVservices in the last decade The high-definition broadcast got
popular since its introduction in 2004 and the accessibility of3D content is growing year by year The consumer electronicdevices become integrated part of our life customers accessdigital content from set-top boxes (STBs) to tablets andmobile devices From the IPTV service provider point ofview the demand of high-quality services emerged howeverthe infrastructure of the access network remained the sameThe main technology of telecommunication operators pro-viding internet remained the 20ndash30 years old twisted copperpairs
On one hand the IPTV Service Providers are motivatedby the maximization of their customer reach but in manycases digital subscriber line xDSL offers inadequate band-width for high quality services [1] access network providersneed to find a solution which enables them to utilize theircurrent infrastructure The answers may include the imple-mentation of a more advanced encoding algorithm (H264H265) which results in having the same quality on smallerthroughput [2] or the introduction of a hybrid service whichreplaces the most bandwidth consuming scheduled contenttransport with digital video broadcasting DVB-X technology[3] the usage of progressive download or as we point out inour work the implementation of a more effective bandwidthallocation in the access network which would ensure a moreefficient transport
2 International Journal of Digital Multimedia Broadcasting
Bandwidth
Prio 3Internet
Prio 2IPTV
Prio 1-voice
SD
SD
SD stream
HD streamRETAV data
Figure 1 Bandwidth allocation in the access network
On the other hand the service portability allows con-sumers to access the content not only on STBs but also fromvarious hand-held devices Today the prime wireless accesstechnology within the consumer domain is the wireless localarea network WLAN connection therefore service platformproviders have to adapt their IPTV solution to supportthe specific requirements of this wireless communicationchannel
Most of the current research papers [4ndash6] take onlyone of the above mentioned two aspects into considerationThe effect of the wireless channel is usually addressed onthe media access control layer and the research of theIPTV distribution focuses only on the quality of the fixednetwork infrastructure Our research does address the abovedescribed overlap of the IPTV delivery over WLAN in a lim-ited resource environment of xDSL technology We presentthe concept and main problems of bandwidth allocationin Section 3 Our new theoretical model is introduced bySection 4 We develop a more effective bandwidth allocationalgorithm in Section 5 which we evaluate and validateFinally Section 6 concludes our work and shows somepotential further application areas
3 Bandwidth Allocation
Let us begin the discussion of bandwidth managementby introducing a typical triple-play bandwidth allocationscheme on Figure 1 Access network providers usually ded-icate a reserved bandwidth in the access network forvoice communication and share the remaining throughputbetween IPTV and internet services with a priority for theformer one
The actual throughput of IPTV service depends on theuserrsquos configuration In most of the cases a token-basedstream management allows the customers to simultaneouslyreceive multiple streams for live viewing or recording pur-poses (1SD+1HD or 3SD+0HD)
One token allocates bandwidth for the AV (audio-video)data transport and reserves a dedicated bandwidth for theretransmission RET service We discuss the problem andtradeoff of this bandwidth allocation therefore in the follow-ing paragraphs we are going to describe it in details
The balance between the assigned bandwidth for AV dataand the reserved bandwidth for RET service is crucial forachieving the maximal quality in IPTV solutions On onehand the stream bandwidth as constant in time (because ofthe widely applied is considered constant bitrate CBR video
Throughput
119875LOSS
119861stream
119861AV data
119861crit
Opportunity-a Opportunity-b
119875overload 119875critstat 119875critdyn
Dynamic BW allocationStatic BW allocation
Figure 2 Static versus dynamic bandwidth allocation
encoding) the more throughput is assigned for the AV dataand the better stream quality can be achieved by the increaseof the encoding bitrate but the less opportunity is given forerror correction A smooth sharp stream may be disturbedby blocking or full frame outages due to the insufficient RETthroughput On the other hands reserving high bandwidthfor error correction degrades the overall stream quality dueto the low encoding bitrate Based on different networkinstallations the ratio of RET bandwidth toAVdata is usuallytuned between 10 and 25 but a suboptimal value maysignificantly reduce the throughput and quality of an IPTVSolution
4 Proposed Model for Bandwidth AllocationIPTV Solutions in WLAN Home Networks
Themain concept and benefits of our research are showed byFigure 2 The solid line represents the theoretical throughputof the AV data stream at various packet loss probabilities incase of static bandwidth allocation The function is constanttill 119875overload loss probability where the loss rate is so highthat the retransmission traffic fully occupies its reservedbandwidth Above this rate retransmission does not haveenough bandwidth to recover all the packet losses thereforethe actual throughput of AV data stream is decreasing(not transmitted packets) and customers experience qualitydeterioration We also declare a critical bandwidth value(119861crit) for AV data transmission [Below this value it does notmake sense to provide IPTV service due to themassive losses
A dynamic bandwidth allocation (dash-dot line onFigure 2) increases the throughput of the AV stream atlow packet loss rates to achieve a better quality for theIPTV service (opportunity-a) Secondly at higher loss ratesthe throughput of AV stream decreased to avoid doublepacket delivery (opportunity-b) which enables the operatorto expand the value of the critical packet loss rate (119875critdyn)and provides a lower quality but error-free IPTV service in aworst environment
In this paper we present only one part of our over-all research the optimal selection of the retransmissionthroughput Our methodmdashintroduced by the upcomingsectionsmdashpredicts the loss attributes of the wireless transmis-sion and defines the optimal value of the RET throughputconsidering the overall loss parameters with the aim of
International Journal of Digital Multimedia Broadcasting 3
Channel119885119899
119885119899
minus1
119861 lowast119861119899
119889
119889 119889
119862119899
++119878119899
119878119899
TX RX119877119899 119877
998400119899
Figure 3 Channel model
minimizing packet losses First we describe the channel andthe bandwidth models
41The ChannelModel Considering the requirements aboveand the attributes of aWLAN transport we decided to intro-duce a discrete-time channel and a Markov model for themathematical description of the UNIT-17 interface (AV datastreams in the Access and residential Networks)We considernot only the packet arrival but also the retransmission trafficas a discrete-time stochastic process and we prove that it is ahomogeneous Markov chain
Let 119878119899
= 0 for all 119899 represent the number of sentpackets at the time 119899 on Figure 3 TX the multicast contentdelivery function119885
119899
= 0 1 for all 119899 an additive white noisein the wireless communication channel d the transmissiondelay and RX the open IPTV terminal function (OITF) Thereceived packets are expressed with 119877
119899
= 0 for successfulpacket transmission 1 for packet loss We assume that thereceiver detects a packet loss (by checking the sequencenumbers of packets) and requests the retransmission119861
119899
= 1for a packet retransmission 0 otherwise of every lost packetsonly once from the B Fast Channel ChangeRetransmissionserver We also assume that the retransmission requestcommunication is protected by an error-free protocol likeTCP By this definition we obtain
119861119899
= 119877119899
= 119885119899minus119889
(1)
The 119861119899
signal travels through the same wireless channeltherefore it is also effected by the same 119885
119899
noise and it maybe also lost We model this effect by expressing the receivedcorrection signal119862
119899
= 0 1with three operators an inverter(minus1) a multiplier (lowast) and a channel transmission delay (d)These functions enable us to assign the value of 1 for 119862
119899
onlyin the case when the retransmission signal is not effected bythe channel noise (not lost) and the value of 0 otherwise
The final received and corrected signal 1198771015840119899
= 0 for suc-cessful packet transmission 1 for packet loss (unsuccessfulretransmission) and 2 for successful packet retransmissionis
1198771015840
119899
= 119877119899minus119889
+ 119862119899
= 119878119899minus2119889
+ 119885119899minus2119889
+ 119861119899minus119889
lowast 119885minus1
119899minus119889
= 119878119899minus2119889
+ 119885119899minus2119889
+ 119885119899minus2119889
lowast 119885minus1
119899minus119889
(2)
Let us observe that the first term of the addition equalsto 0 by definition and the last term equals to the samplingof the 119885
119899
white noise with its own delayed signal Theautocorrelation of the white noise is zero for all nonzero timeshifts [7] therefore 1198771015840
119899
can be described as a sequence ofan independent random variable which satisfies the Markovproperty119883
119899
def= 1198771015840
119899
is a homogeneous Markov chainLet119883119899
= 0 if the 119899th packet is received correctly119883119899
= 1 ifthe 119899th packet is lost and has not been retransmitted finally119883119899
= 2 if the 119899th packet is successfully retransmitted afterloss We analyzed and described a three-state Markov modelin our previous publication therefore we simply list mostimportant properties For a detailed discussion including theresolution of (3)ndash(6) and for themeaning of the probabilitiesplease refer to our former paper [8]
The transition matrix
[
[
1 minus 11990101
minus 11990102
11990101
11990102
11990110
1 minus 11990110
minus 11990112
11990112
11990120
11990121
1 minus 11990120
minus 11990121
]
]
(3)
The steady-state packet loss rate
119875LOSSsteady = (11990102
11990121
+ 11990101
11990120
+ 11990101
11990121
)
times (11990101
11990120
+ 11990101
11990121
+ 11990102
11990121
+ 11990110
11990102
+ 11990112
11990101
+11990112
11990102
+ 11990121
11990110
+ 11990120
11990110
+ 11990120
11990112
)minus1
(4)
The steady-state packet retransmission rate
119875RETsteady = (11990110
11990102
+ 11990112
11990101
+ 11990112
11990102
)
times (11990101
11990120
+ 11990101
11990121
+ 11990102
11990121
+ 11990110
11990102
+ 11990112
11990101
+11990112
11990102
+ 11990121
11990110
+ 11990120
11990110
+ 11990120
11990112
)minus1
(5)
And the probability of loss burst with length 119897
119875RETburst (119897) = (1 minus 11990120
minus 11990121
)119897minus1
(11990120
+ 11990121
) (6)
42TheModel of Bandwidth Limitation Theprevious sectionintroduced how the wireless channel affects the packet trans-mission and now we are going to analyze the the bandwidthallocation in IPTV solutions
We introduce three planes of the IPTV packet trans-mission The transmitter plane represents the providerrsquosnetwork receiver plane represents the OITF and playout planrepresents the content presentation within the OITF Theevents of the packet loss usually show a burstiness in wirelesscommunication [9ndash11] Therefore we investigate an intra-burst loss on Figure 4 after the first 119896 consecutive packetlosses (b) the receiver requests them for retransmission (c)which packets are delivered within the allocated bandwidthfor packet transmission (d) to the presentation device (e) Fora successful retransmission all retransmitted packet should
4 International Journal of Digital Multimedia Broadcasting
Bandwidth
(c) Retransmitted
(a) Sent packets
(d) Ret received
(b) Lost packets
(e) Played out
Transmitter
Receiver
Playout
1 2
1 2
RET
AVdata
AVdata
AVdata
1
middot middot middot
middot middot middot
119896
1 2 middot middot middot 119896
119896
middot middot middot 119896
RET 1 2 middot middot middot 119896
119905
119905
119905119879ret119863ret 119879playo119863playo
Figure 4 Intraburst retransmission
arrive earlier than their presentation time (119879ret le 119879playo)Expressing this condition with the durations
119863pkg + RTT + 119896119861AV119861RET
119863pkg lt (119896 minus 1)119863pkg + 119863playo (7)
where 119863pkg is the average transmission time for a packet119877119879119879 is the round-trip time 119861AV and 119861RET are the allocatedbandwidths on the communication channel and 119863playo isthe packet playout buffer in the OITF The formula ofthe maximal number of consecutive packets which can besuccessfully retransmitted is defined as
119896RETintramax (119861RET119861AV
) =
119863playo minus 2119863pkg minus RTT119863pkg ((1 (119861RET119861AV)) minus 1)
(8)
The actual throughput of the AV stream may vary byinstallations therefore we expressed this value as a ratio of119861RET and 119861AV
Second we highlight the barrier of interburst behavioron Figure 5 After a loss of long burst the retransmissionbandwidth is occupied by the traffic of the retransmittedpackets even if there is no other packet loss at the time inthe video stream This means that a loss event blocks theretransmission channel We are interested in the followingquestion assuming a 119896 lt 119896RETintramax long burst of loss afterhowmany packets (119899) can a new loss burst occurwhichwouldbe also successfully retransmitted (eg what is the minimaldistance (119899 minus 119896) between two loss bursts if the first burst lastsfor 119896 packets) Now 119879ret119896
def= 119879playo119896 and 119879ret119899 le 119879playo119899
Figure 5 shows that
119863pkg + RTT + (119896 + 1)119861AV119861RET
119863pkg lt (119899 minus 1)119863pkg + 119863playo
(9)
where 119899 gt 119896 Expressing 119899
119899RETintermin (119896119861AV119861RET
) = (119896 + 1)119861AV119861RET
+
RTT minus 119863playo
119863pkg+ 2
(10)
119879playo 119899
119879ret119899119879ret119896119863ret119896
Bandwidth
(c) Retransmitted
(a) Sent packets
(d) Ret received
(b) Lost packets (e) Played out
Transmitter
Receiver
Playout
RET
AVdata
AVdata
AVdata
middot middot middot
middot middot middot middot middot middot middot middot middot middot middot middot
middot middot middot middot middot middot
middot middot middot middot middot middot middot middot middot middot middot middot
middot middot middot
1 119896
RET 11
middot middot middot
119896119896
119905
119905
119905
119863playo
119899
1 middot middot middot 119896 119899
119899
119899
1 119896 119899
middot middot middot middot middot middot middot middot middot
middot middot middotmiddot middot middotmiddot middot middot
middot middot middot
Figure 5 Interburst retransmission
0
20
40
60
80
100
120
0 005 01 015 02 025 03 035 04
Num
ber o
f pac
kets
119861RET 119861AV
119899RETintermin 119863playo = 1000ms119896RETintramax 119863playo = 1000ms119899RETintermin 119863playo = 800ms119896RETintramax 119863playo = 800ms119899RETintermin 119863playo = 600ms119896RETintramax 119863playo = 600ms119899RETintermin 119863playo = 400ms119896RETintramax 119863playo = 400ms119899RETintermin 119863playo = 200ms119896RETintramax 119863playo = 200ms
Figure 6 Intra- and interburst limitation
We present the above declared functions (8)ndash(10) onFigure 6 This graph shows that at small values of 119861AV119861RET(0ndash15) the effect of the inter-burst blocking is greaterFor example at 5 the maximal consecutive burst lengthis 10 packets and the retransmission channel is blocked bythis traffic for 30 packets Above 20 this effect becomesinsignificant
We also state that with grater playout buffer (119863playo) RETis able to correct larger loss bursts at the price of a greater end-to-end delay
International Journal of Digital Multimedia Broadcasting 5
5 The Optimal Retransmission
In this sectionwe introduce a newmethod for retransmissionbandwidth allocation based on our models with the aimof achieving a better video quality The method is realizedin a test environment and our hypothesis is measured andproved
Traditional RET algorithms request all lost packets there-fore they have to implement a network layer traffic shapingto fit the actual retransmission throughput into the allocatedbandwidth This is usually done by packet queuing whichincreases the overall packet retransmission time thereforethe probability of a retransmission packet arrives late (afterits playout time) is great Several papers addressed thisproblem [12] introducing a selective retransmission protocolby evaluating the traffic on the application level and assigningpriority and importance for each packet retransmissions
The main advantages of our method are the minimaladditional delay the low resource needs and the consid-eration of the wireless channel Our method assess theRET mechanism on the network layer skips (forbids) theretransmission requests of a lost packet according to theabove described intra- and interburst channel blocking andtakes the special properties of the wireless channel intoconsideration Our algorithm consists of three steps
(1) The packet arrival process is continuously monitoredfor packet loss
(2) 119896RETintramax and 119899RETintermin are calculated(3) A lost packet is requested retransmission only if the
intra- and inter-burst channel blocking do not for-bid the retransmission otherwise the retransmissionrequest is skipped
51 Empirical Evaluation To evaluate our model and meth-ods we followed theOIPF system architecture [13] and imple-mented the following OIPF functions in a test environment(source codes are available on [14])
(1) Multicast content delivery function ser c++ applica-tion generates UDPRTP multicast traffic and imple-ments a simple control protocol hosted by an x86Linux server connected to the core network ofDeutsche Telekom (DT)
(2) RET server ret c++ application stores the multicasttraffic in a circular buffer and implements a simpleretransmission request protocol hosted by the samex86 Linux server
(3) Unit-17 interface part a was realized by the ADSL2+access network of DT and was provided by a DLinkADSL modem
(4) Unit-17 interface part b was realized by a 80211 bWLAN network and was provided by a Cisco 1200series wireless access point (AP) connected to theADSL modem
(5) OITF cli c++ application implements a simple mul-ticast receiver and controls functions of the multicast
0
50
100
150
200
0 50 100 150 200
Thro
ughp
ut (k
bps)
Time (s)Reserved bandwidthRET allRET skip
Figure 7 Retransmission throughput
content delivery function and RET server Hosted ona x86 Linux laptop connected to the Cisco AP
Figure 7 shows the throughput of the retransmissionstream in two cases traditional retransmission (RET all alllost packets were requested for retransmission) and our newretransmission method (RET skip retransmission requestsmay be skipped based on the actual parameters of thechannel) It can be clearly seen that our algorithm kept theretransmission throughput under its dedicated bandwidthwhich ensured the in-time delivery of the retransmissionpackets however we intentionally skipped those retransmis-sion requests which in time delivery would not be ensureddue to the channel blocking (intra- and inter-burst effect)
Themain benefit of ourmethod is showed by Figure 8Wecompared the total packet loss rate in the above mentionedtwo cases and we found that with the smart skipping ofpacket retransmission requests we were able to achievea better (smaller) loss rate then retransmitting all of thepacketsOurmethod avoided the effect of late retransmissionIF a packet is requested for retransmission without ensuringthe necessary transport bandwidth then it may delay furtherretransmission requests which may arrive to late after theirplayout time This causes an inefficient retransmission band-width utilization which increases the overall packet loss rate(on the playout plane)
52 The Effect of the Intraburst Limitation Let us analyzeour results theoretically as well In this and in the upcomingsection we characterize the Unit-17 interface with the transi-tion matrix of our Markov model and the design attributesof the access network Applying the intra- and inter-burstlimitations on our model we derive the probability of theretransmission skip caused by our algorithm Finally weexpress the overall packet loss rate which is a key indicatorfor the quality of the video transmission
6 International Journal of Digital Multimedia Broadcasting
0
001
002
003
004
005
006
007
008
009
01
0 50 100 150 200
Ove
rall
pack
et lo
ss ra
te
Time (s)
RET allRET skip
Figure 8 Overall packet loss rate
The intra-burst limitations have a significant short-timeeffect if the distance of the burst losses is great (119901
00
rarr
1) Our question is about the probability of retransmissionpacket skip First we calculate the maximal number ofconsecutive retransmission requests
119896RETintramax =
119863playo minus 2119863pkg minus RTT119863pkg ((119861AV119861RET) minus 1)
(11)
The probability of 119897 long retransmission request is givenby the Markov model (6) We calculate the probability of119898 packet skips if the retransmission burst is greater than119896119877burstmax which is
119875skip (119898) = (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1+119898
(12)
The overall probability of a packet skip is given by
119875skipintra = lim119894rarrinfin
infin
sum
119894=1
119875skip (119894)
119894
= lim119894rarrinfin
infin
sum
119894=1
(11990120
+ 11990121
) (1minus 11990120
minus 11987521
)119896RETintramaxminus1+119894
119894
= (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1
times lim119894rarrinfin
infin
sum
119894=1
(1 minus 11990120
minus 11990121
)119894
119894
(13)
Let us observe that the last sum can be expressed as aspecial form of the polylogarithm (also known as Jonquierersquosfunction)
119871119894119904
(119911) |119904=1
=
infin
sum
119896=1
119911119896
119896119904|119904=1
=
infin
sum
119896=1
119911119896
119896 (14)
for every minus1 le 119911 lt 1 (1 minus 11990120
minus 11987521
satisfies this criteriaTherefore using the well-known formula of 119871119894
1
(119911) = minus ln(1 minus
119911) the equation can be expressed in a closed form
119875skipintra = (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1
times (minus1) ln (1 minus (1 minus 11990120
minus 11990121
))
= (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1
times (minus1) ln (11990120
+ 11990121
)
(15)
The overall packet loss can be expressed as a sum of theprobability of packet skip (15) and the steady state probabilityof packet loss (4)
53 The Effect of the Interburst Limitation We ask the samequestion as in the previous section what is the probability ofpacket skip Let us assume that the first burst is small enoughto be retransmitted (119896 lt 119896RETintramax) The probability ofretransmission burst is of 119897 size is given by ourMarkov model(6)The probability of 119896 retransmission burst followed by 119899minus119896good transmission burst and a second retransmission
119891 (119896 119899) = (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896minus1
11990120
times (11990101
+ 11990102
)
times (1 minus 11990101
minus 11990102
)119899minus119896minus1
11990102
(16)
The first packet of the second retransmission burst isskipped if 119899 lt 119899RETintermin(119896) From this we can calculatethe probability of one packet skip for 119896
119875skipinterk =
119899RETintermin(119896)
sum
119894=119896
119891 (119896 119894) (17)
For the overal packet skip probability we have to summa-rize (17) for every 119896 lt 119896Rburstmax
119875skipinter =
119896RETintramax
sum
119895=1
119899RETintermin(119895)
sum
119894=119895
119891 (119895 119894) (18)
The overall packet loss can be expressed as a sum of theprobability of packet skip (18) and the steady-state probabilityof packet loss (4)
6 Conclusion
In this paper we highlighted the relevance of the wirelesstransport in todayrsquos IPTV solutions and we pointed out thatitrsquos and the access networkrsquos combined effect on the band-width management is not discussed deeply by publicationsOur general research project targets this specific area byintroducing several optimization methods from which wepresented one the optimization of the packet retransmissionon the previous pages
We created a newdiscrete-time channelmodel to describethe effect of the burst losses on the IPTV service quality and
International Journal of Digital Multimedia Broadcasting 7
the role of the packet retransmission We proved that it isa Markov Chain and as part of our results we expressedand evaluated its most important quantitative parametersUsing our model we also introduced an algorithm forretransmission optimization in IPTV solutions over WLANhome networks
As a further evaluation of our results we created atestbed in alignment with the OPIF system architecture andwe performed the empirical analysis of our channel modeland methods We showed that our concept improved theoverall packet loss characteristics Furthermore we encloseda mathematical analysis of our algorithm and we derived thetheoretical packet loss probabilities to support our measure-ments
In the next research phases we are going to investigateintroduce and leverage our theoretical results of throughputmanagement in the adaptive bitrate streaming technologiesfor IPTV solutions and we are going to evaluate our channelmodel in the media access control layer of the wireless accesstechnologies
References
[1] Open IPTV Forum ldquoRelease 2 Specificationsrdquo httpwwwoipftvspecificationshtml
[2] A Abramowski ldquoTowards H 265 video coding standardrdquo inProceedings of the Photonics Applications in Astronomy Commu-nications Industry and High-Energy Physics Experiments vol8008 of Proceedings of the SPIE 2011
[3] H Benot Digital Television Satellite Cable Terrestrial IptvMobile Tv in the Dvb Framework Focal Press 2008
[4] MMHassani S V Jalali andAAkbari ldquoEvaluatingTCPflowsbehavior by a mathematical model in WLANrdquo Journal of BasicandApplied Scientific Research vol 2 no 3 pp 2809ndash2814 2012
[5] A C Begen ldquoError control for IPTV over xDSL networksrdquoin Proceedings of the 5th IEEE Consumer Communications andNetworking Conference (CCNC rsquo08) pp 632ndash637 January 2008
[6] H Bobarshad M Van Der Schaar and M R Shikh-BahaeildquoA low-complexity analytical modeling for cross-layer adaptiveerror protection in video over WLANrdquo IEEE Transactions onMultimedia vol 12 no 5 pp 427ndash438 2010
[7] A Papoulis Probability Random Variables and Stohastic Pro-cesses McGraw-Hill 1965
[8] T Jursonovics and S Imre ldquoAnalysis of a newMarkovmodel forpacket loss characterization in IPTV Solutionsrdquo Infocommuni-cations Journal vol 2 pp 28ndash33 2011
[9] T Jursonovics and Zs Butyka ldquoThe implementation and analy-sis of interactive multimedia mobile servicesrdquo in Proceedings ofthe 11th Microcoll Conference pp 149ndash152 Budapest Hungary2003
[10] T Jursonovics Zs Butyka and S Imre ldquoMultimedia transmis-sion over mobile networksrdquo in International Conference on Soft-ware Telecommunications and Computer Networks (SoftCOMrsquo04) pp 453ndash457 Dubrovnik Croatia Ancona Italy 2004
[11] A Nafaa T Taleb and L Murphy ldquoForward error correctionstrategies for media streaming over wireless networksrdquo IEEECommunications Magazine vol 46 no 1 pp 72ndash79 2008
[12] A Huszak and S Imre ldquoSource controlled semi-reliable mul-timedia streaming using selective retransmission in DCCPIP
networksrdquo Computer Communications vol 31 no 11 pp 2676ndash2684 2008
[13] Open IPTV Forum Functional Architecture [V2 1] [2011-03-15] httpwwwoipftvspecificationshtml
[14] httpwwwmclhusimjursonovics
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
2 International Journal of Digital Multimedia Broadcasting
Bandwidth
Prio 3Internet
Prio 2IPTV
Prio 1-voice
SD
SD
SD stream
HD streamRETAV data
Figure 1 Bandwidth allocation in the access network
On the other hand the service portability allows con-sumers to access the content not only on STBs but also fromvarious hand-held devices Today the prime wireless accesstechnology within the consumer domain is the wireless localarea network WLAN connection therefore service platformproviders have to adapt their IPTV solution to supportthe specific requirements of this wireless communicationchannel
Most of the current research papers [4ndash6] take onlyone of the above mentioned two aspects into considerationThe effect of the wireless channel is usually addressed onthe media access control layer and the research of theIPTV distribution focuses only on the quality of the fixednetwork infrastructure Our research does address the abovedescribed overlap of the IPTV delivery over WLAN in a lim-ited resource environment of xDSL technology We presentthe concept and main problems of bandwidth allocationin Section 3 Our new theoretical model is introduced bySection 4 We develop a more effective bandwidth allocationalgorithm in Section 5 which we evaluate and validateFinally Section 6 concludes our work and shows somepotential further application areas
3 Bandwidth Allocation
Let us begin the discussion of bandwidth managementby introducing a typical triple-play bandwidth allocationscheme on Figure 1 Access network providers usually ded-icate a reserved bandwidth in the access network forvoice communication and share the remaining throughputbetween IPTV and internet services with a priority for theformer one
The actual throughput of IPTV service depends on theuserrsquos configuration In most of the cases a token-basedstream management allows the customers to simultaneouslyreceive multiple streams for live viewing or recording pur-poses (1SD+1HD or 3SD+0HD)
One token allocates bandwidth for the AV (audio-video)data transport and reserves a dedicated bandwidth for theretransmission RET service We discuss the problem andtradeoff of this bandwidth allocation therefore in the follow-ing paragraphs we are going to describe it in details
The balance between the assigned bandwidth for AV dataand the reserved bandwidth for RET service is crucial forachieving the maximal quality in IPTV solutions On onehand the stream bandwidth as constant in time (because ofthe widely applied is considered constant bitrate CBR video
Throughput
119875LOSS
119861stream
119861AV data
119861crit
Opportunity-a Opportunity-b
119875overload 119875critstat 119875critdyn
Dynamic BW allocationStatic BW allocation
Figure 2 Static versus dynamic bandwidth allocation
encoding) the more throughput is assigned for the AV dataand the better stream quality can be achieved by the increaseof the encoding bitrate but the less opportunity is given forerror correction A smooth sharp stream may be disturbedby blocking or full frame outages due to the insufficient RETthroughput On the other hands reserving high bandwidthfor error correction degrades the overall stream quality dueto the low encoding bitrate Based on different networkinstallations the ratio of RET bandwidth toAVdata is usuallytuned between 10 and 25 but a suboptimal value maysignificantly reduce the throughput and quality of an IPTVSolution
4 Proposed Model for Bandwidth AllocationIPTV Solutions in WLAN Home Networks
Themain concept and benefits of our research are showed byFigure 2 The solid line represents the theoretical throughputof the AV data stream at various packet loss probabilities incase of static bandwidth allocation The function is constanttill 119875overload loss probability where the loss rate is so highthat the retransmission traffic fully occupies its reservedbandwidth Above this rate retransmission does not haveenough bandwidth to recover all the packet losses thereforethe actual throughput of AV data stream is decreasing(not transmitted packets) and customers experience qualitydeterioration We also declare a critical bandwidth value(119861crit) for AV data transmission [Below this value it does notmake sense to provide IPTV service due to themassive losses
A dynamic bandwidth allocation (dash-dot line onFigure 2) increases the throughput of the AV stream atlow packet loss rates to achieve a better quality for theIPTV service (opportunity-a) Secondly at higher loss ratesthe throughput of AV stream decreased to avoid doublepacket delivery (opportunity-b) which enables the operatorto expand the value of the critical packet loss rate (119875critdyn)and provides a lower quality but error-free IPTV service in aworst environment
In this paper we present only one part of our over-all research the optimal selection of the retransmissionthroughput Our methodmdashintroduced by the upcomingsectionsmdashpredicts the loss attributes of the wireless transmis-sion and defines the optimal value of the RET throughputconsidering the overall loss parameters with the aim of
International Journal of Digital Multimedia Broadcasting 3
Channel119885119899
119885119899
minus1
119861 lowast119861119899
119889
119889 119889
119862119899
++119878119899
119878119899
TX RX119877119899 119877
998400119899
Figure 3 Channel model
minimizing packet losses First we describe the channel andthe bandwidth models
41The ChannelModel Considering the requirements aboveand the attributes of aWLAN transport we decided to intro-duce a discrete-time channel and a Markov model for themathematical description of the UNIT-17 interface (AV datastreams in the Access and residential Networks)We considernot only the packet arrival but also the retransmission trafficas a discrete-time stochastic process and we prove that it is ahomogeneous Markov chain
Let 119878119899
= 0 for all 119899 represent the number of sentpackets at the time 119899 on Figure 3 TX the multicast contentdelivery function119885
119899
= 0 1 for all 119899 an additive white noisein the wireless communication channel d the transmissiondelay and RX the open IPTV terminal function (OITF) Thereceived packets are expressed with 119877
119899
= 0 for successfulpacket transmission 1 for packet loss We assume that thereceiver detects a packet loss (by checking the sequencenumbers of packets) and requests the retransmission119861
119899
= 1for a packet retransmission 0 otherwise of every lost packetsonly once from the B Fast Channel ChangeRetransmissionserver We also assume that the retransmission requestcommunication is protected by an error-free protocol likeTCP By this definition we obtain
119861119899
= 119877119899
= 119885119899minus119889
(1)
The 119861119899
signal travels through the same wireless channeltherefore it is also effected by the same 119885
119899
noise and it maybe also lost We model this effect by expressing the receivedcorrection signal119862
119899
= 0 1with three operators an inverter(minus1) a multiplier (lowast) and a channel transmission delay (d)These functions enable us to assign the value of 1 for 119862
119899
onlyin the case when the retransmission signal is not effected bythe channel noise (not lost) and the value of 0 otherwise
The final received and corrected signal 1198771015840119899
= 0 for suc-cessful packet transmission 1 for packet loss (unsuccessfulretransmission) and 2 for successful packet retransmissionis
1198771015840
119899
= 119877119899minus119889
+ 119862119899
= 119878119899minus2119889
+ 119885119899minus2119889
+ 119861119899minus119889
lowast 119885minus1
119899minus119889
= 119878119899minus2119889
+ 119885119899minus2119889
+ 119885119899minus2119889
lowast 119885minus1
119899minus119889
(2)
Let us observe that the first term of the addition equalsto 0 by definition and the last term equals to the samplingof the 119885
119899
white noise with its own delayed signal Theautocorrelation of the white noise is zero for all nonzero timeshifts [7] therefore 1198771015840
119899
can be described as a sequence ofan independent random variable which satisfies the Markovproperty119883
119899
def= 1198771015840
119899
is a homogeneous Markov chainLet119883119899
= 0 if the 119899th packet is received correctly119883119899
= 1 ifthe 119899th packet is lost and has not been retransmitted finally119883119899
= 2 if the 119899th packet is successfully retransmitted afterloss We analyzed and described a three-state Markov modelin our previous publication therefore we simply list mostimportant properties For a detailed discussion including theresolution of (3)ndash(6) and for themeaning of the probabilitiesplease refer to our former paper [8]
The transition matrix
[
[
1 minus 11990101
minus 11990102
11990101
11990102
11990110
1 minus 11990110
minus 11990112
11990112
11990120
11990121
1 minus 11990120
minus 11990121
]
]
(3)
The steady-state packet loss rate
119875LOSSsteady = (11990102
11990121
+ 11990101
11990120
+ 11990101
11990121
)
times (11990101
11990120
+ 11990101
11990121
+ 11990102
11990121
+ 11990110
11990102
+ 11990112
11990101
+11990112
11990102
+ 11990121
11990110
+ 11990120
11990110
+ 11990120
11990112
)minus1
(4)
The steady-state packet retransmission rate
119875RETsteady = (11990110
11990102
+ 11990112
11990101
+ 11990112
11990102
)
times (11990101
11990120
+ 11990101
11990121
+ 11990102
11990121
+ 11990110
11990102
+ 11990112
11990101
+11990112
11990102
+ 11990121
11990110
+ 11990120
11990110
+ 11990120
11990112
)minus1
(5)
And the probability of loss burst with length 119897
119875RETburst (119897) = (1 minus 11990120
minus 11990121
)119897minus1
(11990120
+ 11990121
) (6)
42TheModel of Bandwidth Limitation Theprevious sectionintroduced how the wireless channel affects the packet trans-mission and now we are going to analyze the the bandwidthallocation in IPTV solutions
We introduce three planes of the IPTV packet trans-mission The transmitter plane represents the providerrsquosnetwork receiver plane represents the OITF and playout planrepresents the content presentation within the OITF Theevents of the packet loss usually show a burstiness in wirelesscommunication [9ndash11] Therefore we investigate an intra-burst loss on Figure 4 after the first 119896 consecutive packetlosses (b) the receiver requests them for retransmission (c)which packets are delivered within the allocated bandwidthfor packet transmission (d) to the presentation device (e) Fora successful retransmission all retransmitted packet should
4 International Journal of Digital Multimedia Broadcasting
Bandwidth
(c) Retransmitted
(a) Sent packets
(d) Ret received
(b) Lost packets
(e) Played out
Transmitter
Receiver
Playout
1 2
1 2
RET
AVdata
AVdata
AVdata
1
middot middot middot
middot middot middot
119896
1 2 middot middot middot 119896
119896
middot middot middot 119896
RET 1 2 middot middot middot 119896
119905
119905
119905119879ret119863ret 119879playo119863playo
Figure 4 Intraburst retransmission
arrive earlier than their presentation time (119879ret le 119879playo)Expressing this condition with the durations
119863pkg + RTT + 119896119861AV119861RET
119863pkg lt (119896 minus 1)119863pkg + 119863playo (7)
where 119863pkg is the average transmission time for a packet119877119879119879 is the round-trip time 119861AV and 119861RET are the allocatedbandwidths on the communication channel and 119863playo isthe packet playout buffer in the OITF The formula ofthe maximal number of consecutive packets which can besuccessfully retransmitted is defined as
119896RETintramax (119861RET119861AV
) =
119863playo minus 2119863pkg minus RTT119863pkg ((1 (119861RET119861AV)) minus 1)
(8)
The actual throughput of the AV stream may vary byinstallations therefore we expressed this value as a ratio of119861RET and 119861AV
Second we highlight the barrier of interburst behavioron Figure 5 After a loss of long burst the retransmissionbandwidth is occupied by the traffic of the retransmittedpackets even if there is no other packet loss at the time inthe video stream This means that a loss event blocks theretransmission channel We are interested in the followingquestion assuming a 119896 lt 119896RETintramax long burst of loss afterhowmany packets (119899) can a new loss burst occurwhichwouldbe also successfully retransmitted (eg what is the minimaldistance (119899 minus 119896) between two loss bursts if the first burst lastsfor 119896 packets) Now 119879ret119896
def= 119879playo119896 and 119879ret119899 le 119879playo119899
Figure 5 shows that
119863pkg + RTT + (119896 + 1)119861AV119861RET
119863pkg lt (119899 minus 1)119863pkg + 119863playo
(9)
where 119899 gt 119896 Expressing 119899
119899RETintermin (119896119861AV119861RET
) = (119896 + 1)119861AV119861RET
+
RTT minus 119863playo
119863pkg+ 2
(10)
119879playo 119899
119879ret119899119879ret119896119863ret119896
Bandwidth
(c) Retransmitted
(a) Sent packets
(d) Ret received
(b) Lost packets (e) Played out
Transmitter
Receiver
Playout
RET
AVdata
AVdata
AVdata
middot middot middot
middot middot middot middot middot middot middot middot middot middot middot middot
middot middot middot middot middot middot
middot middot middot middot middot middot middot middot middot middot middot middot
middot middot middot
1 119896
RET 11
middot middot middot
119896119896
119905
119905
119905
119863playo
119899
1 middot middot middot 119896 119899
119899
119899
1 119896 119899
middot middot middot middot middot middot middot middot middot
middot middot middotmiddot middot middotmiddot middot middot
middot middot middot
Figure 5 Interburst retransmission
0
20
40
60
80
100
120
0 005 01 015 02 025 03 035 04
Num
ber o
f pac
kets
119861RET 119861AV
119899RETintermin 119863playo = 1000ms119896RETintramax 119863playo = 1000ms119899RETintermin 119863playo = 800ms119896RETintramax 119863playo = 800ms119899RETintermin 119863playo = 600ms119896RETintramax 119863playo = 600ms119899RETintermin 119863playo = 400ms119896RETintramax 119863playo = 400ms119899RETintermin 119863playo = 200ms119896RETintramax 119863playo = 200ms
Figure 6 Intra- and interburst limitation
We present the above declared functions (8)ndash(10) onFigure 6 This graph shows that at small values of 119861AV119861RET(0ndash15) the effect of the inter-burst blocking is greaterFor example at 5 the maximal consecutive burst lengthis 10 packets and the retransmission channel is blocked bythis traffic for 30 packets Above 20 this effect becomesinsignificant
We also state that with grater playout buffer (119863playo) RETis able to correct larger loss bursts at the price of a greater end-to-end delay
International Journal of Digital Multimedia Broadcasting 5
5 The Optimal Retransmission
In this sectionwe introduce a newmethod for retransmissionbandwidth allocation based on our models with the aimof achieving a better video quality The method is realizedin a test environment and our hypothesis is measured andproved
Traditional RET algorithms request all lost packets there-fore they have to implement a network layer traffic shapingto fit the actual retransmission throughput into the allocatedbandwidth This is usually done by packet queuing whichincreases the overall packet retransmission time thereforethe probability of a retransmission packet arrives late (afterits playout time) is great Several papers addressed thisproblem [12] introducing a selective retransmission protocolby evaluating the traffic on the application level and assigningpriority and importance for each packet retransmissions
The main advantages of our method are the minimaladditional delay the low resource needs and the consid-eration of the wireless channel Our method assess theRET mechanism on the network layer skips (forbids) theretransmission requests of a lost packet according to theabove described intra- and interburst channel blocking andtakes the special properties of the wireless channel intoconsideration Our algorithm consists of three steps
(1) The packet arrival process is continuously monitoredfor packet loss
(2) 119896RETintramax and 119899RETintermin are calculated(3) A lost packet is requested retransmission only if the
intra- and inter-burst channel blocking do not for-bid the retransmission otherwise the retransmissionrequest is skipped
51 Empirical Evaluation To evaluate our model and meth-ods we followed theOIPF system architecture [13] and imple-mented the following OIPF functions in a test environment(source codes are available on [14])
(1) Multicast content delivery function ser c++ applica-tion generates UDPRTP multicast traffic and imple-ments a simple control protocol hosted by an x86Linux server connected to the core network ofDeutsche Telekom (DT)
(2) RET server ret c++ application stores the multicasttraffic in a circular buffer and implements a simpleretransmission request protocol hosted by the samex86 Linux server
(3) Unit-17 interface part a was realized by the ADSL2+access network of DT and was provided by a DLinkADSL modem
(4) Unit-17 interface part b was realized by a 80211 bWLAN network and was provided by a Cisco 1200series wireless access point (AP) connected to theADSL modem
(5) OITF cli c++ application implements a simple mul-ticast receiver and controls functions of the multicast
0
50
100
150
200
0 50 100 150 200
Thro
ughp
ut (k
bps)
Time (s)Reserved bandwidthRET allRET skip
Figure 7 Retransmission throughput
content delivery function and RET server Hosted ona x86 Linux laptop connected to the Cisco AP
Figure 7 shows the throughput of the retransmissionstream in two cases traditional retransmission (RET all alllost packets were requested for retransmission) and our newretransmission method (RET skip retransmission requestsmay be skipped based on the actual parameters of thechannel) It can be clearly seen that our algorithm kept theretransmission throughput under its dedicated bandwidthwhich ensured the in-time delivery of the retransmissionpackets however we intentionally skipped those retransmis-sion requests which in time delivery would not be ensureddue to the channel blocking (intra- and inter-burst effect)
Themain benefit of ourmethod is showed by Figure 8Wecompared the total packet loss rate in the above mentionedtwo cases and we found that with the smart skipping ofpacket retransmission requests we were able to achievea better (smaller) loss rate then retransmitting all of thepacketsOurmethod avoided the effect of late retransmissionIF a packet is requested for retransmission without ensuringthe necessary transport bandwidth then it may delay furtherretransmission requests which may arrive to late after theirplayout time This causes an inefficient retransmission band-width utilization which increases the overall packet loss rate(on the playout plane)
52 The Effect of the Intraburst Limitation Let us analyzeour results theoretically as well In this and in the upcomingsection we characterize the Unit-17 interface with the transi-tion matrix of our Markov model and the design attributesof the access network Applying the intra- and inter-burstlimitations on our model we derive the probability of theretransmission skip caused by our algorithm Finally weexpress the overall packet loss rate which is a key indicatorfor the quality of the video transmission
6 International Journal of Digital Multimedia Broadcasting
0
001
002
003
004
005
006
007
008
009
01
0 50 100 150 200
Ove
rall
pack
et lo
ss ra
te
Time (s)
RET allRET skip
Figure 8 Overall packet loss rate
The intra-burst limitations have a significant short-timeeffect if the distance of the burst losses is great (119901
00
rarr
1) Our question is about the probability of retransmissionpacket skip First we calculate the maximal number ofconsecutive retransmission requests
119896RETintramax =
119863playo minus 2119863pkg minus RTT119863pkg ((119861AV119861RET) minus 1)
(11)
The probability of 119897 long retransmission request is givenby the Markov model (6) We calculate the probability of119898 packet skips if the retransmission burst is greater than119896119877burstmax which is
119875skip (119898) = (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1+119898
(12)
The overall probability of a packet skip is given by
119875skipintra = lim119894rarrinfin
infin
sum
119894=1
119875skip (119894)
119894
= lim119894rarrinfin
infin
sum
119894=1
(11990120
+ 11990121
) (1minus 11990120
minus 11987521
)119896RETintramaxminus1+119894
119894
= (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1
times lim119894rarrinfin
infin
sum
119894=1
(1 minus 11990120
minus 11990121
)119894
119894
(13)
Let us observe that the last sum can be expressed as aspecial form of the polylogarithm (also known as Jonquierersquosfunction)
119871119894119904
(119911) |119904=1
=
infin
sum
119896=1
119911119896
119896119904|119904=1
=
infin
sum
119896=1
119911119896
119896 (14)
for every minus1 le 119911 lt 1 (1 minus 11990120
minus 11987521
satisfies this criteriaTherefore using the well-known formula of 119871119894
1
(119911) = minus ln(1 minus
119911) the equation can be expressed in a closed form
119875skipintra = (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1
times (minus1) ln (1 minus (1 minus 11990120
minus 11990121
))
= (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1
times (minus1) ln (11990120
+ 11990121
)
(15)
The overall packet loss can be expressed as a sum of theprobability of packet skip (15) and the steady state probabilityof packet loss (4)
53 The Effect of the Interburst Limitation We ask the samequestion as in the previous section what is the probability ofpacket skip Let us assume that the first burst is small enoughto be retransmitted (119896 lt 119896RETintramax) The probability ofretransmission burst is of 119897 size is given by ourMarkov model(6)The probability of 119896 retransmission burst followed by 119899minus119896good transmission burst and a second retransmission
119891 (119896 119899) = (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896minus1
11990120
times (11990101
+ 11990102
)
times (1 minus 11990101
minus 11990102
)119899minus119896minus1
11990102
(16)
The first packet of the second retransmission burst isskipped if 119899 lt 119899RETintermin(119896) From this we can calculatethe probability of one packet skip for 119896
119875skipinterk =
119899RETintermin(119896)
sum
119894=119896
119891 (119896 119894) (17)
For the overal packet skip probability we have to summa-rize (17) for every 119896 lt 119896Rburstmax
119875skipinter =
119896RETintramax
sum
119895=1
119899RETintermin(119895)
sum
119894=119895
119891 (119895 119894) (18)
The overall packet loss can be expressed as a sum of theprobability of packet skip (18) and the steady-state probabilityof packet loss (4)
6 Conclusion
In this paper we highlighted the relevance of the wirelesstransport in todayrsquos IPTV solutions and we pointed out thatitrsquos and the access networkrsquos combined effect on the band-width management is not discussed deeply by publicationsOur general research project targets this specific area byintroducing several optimization methods from which wepresented one the optimization of the packet retransmissionon the previous pages
We created a newdiscrete-time channelmodel to describethe effect of the burst losses on the IPTV service quality and
International Journal of Digital Multimedia Broadcasting 7
the role of the packet retransmission We proved that it isa Markov Chain and as part of our results we expressedand evaluated its most important quantitative parametersUsing our model we also introduced an algorithm forretransmission optimization in IPTV solutions over WLANhome networks
As a further evaluation of our results we created atestbed in alignment with the OPIF system architecture andwe performed the empirical analysis of our channel modeland methods We showed that our concept improved theoverall packet loss characteristics Furthermore we encloseda mathematical analysis of our algorithm and we derived thetheoretical packet loss probabilities to support our measure-ments
In the next research phases we are going to investigateintroduce and leverage our theoretical results of throughputmanagement in the adaptive bitrate streaming technologiesfor IPTV solutions and we are going to evaluate our channelmodel in the media access control layer of the wireless accesstechnologies
References
[1] Open IPTV Forum ldquoRelease 2 Specificationsrdquo httpwwwoipftvspecificationshtml
[2] A Abramowski ldquoTowards H 265 video coding standardrdquo inProceedings of the Photonics Applications in Astronomy Commu-nications Industry and High-Energy Physics Experiments vol8008 of Proceedings of the SPIE 2011
[3] H Benot Digital Television Satellite Cable Terrestrial IptvMobile Tv in the Dvb Framework Focal Press 2008
[4] MMHassani S V Jalali andAAkbari ldquoEvaluatingTCPflowsbehavior by a mathematical model in WLANrdquo Journal of BasicandApplied Scientific Research vol 2 no 3 pp 2809ndash2814 2012
[5] A C Begen ldquoError control for IPTV over xDSL networksrdquoin Proceedings of the 5th IEEE Consumer Communications andNetworking Conference (CCNC rsquo08) pp 632ndash637 January 2008
[6] H Bobarshad M Van Der Schaar and M R Shikh-BahaeildquoA low-complexity analytical modeling for cross-layer adaptiveerror protection in video over WLANrdquo IEEE Transactions onMultimedia vol 12 no 5 pp 427ndash438 2010
[7] A Papoulis Probability Random Variables and Stohastic Pro-cesses McGraw-Hill 1965
[8] T Jursonovics and S Imre ldquoAnalysis of a newMarkovmodel forpacket loss characterization in IPTV Solutionsrdquo Infocommuni-cations Journal vol 2 pp 28ndash33 2011
[9] T Jursonovics and Zs Butyka ldquoThe implementation and analy-sis of interactive multimedia mobile servicesrdquo in Proceedings ofthe 11th Microcoll Conference pp 149ndash152 Budapest Hungary2003
[10] T Jursonovics Zs Butyka and S Imre ldquoMultimedia transmis-sion over mobile networksrdquo in International Conference on Soft-ware Telecommunications and Computer Networks (SoftCOMrsquo04) pp 453ndash457 Dubrovnik Croatia Ancona Italy 2004
[11] A Nafaa T Taleb and L Murphy ldquoForward error correctionstrategies for media streaming over wireless networksrdquo IEEECommunications Magazine vol 46 no 1 pp 72ndash79 2008
[12] A Huszak and S Imre ldquoSource controlled semi-reliable mul-timedia streaming using selective retransmission in DCCPIP
networksrdquo Computer Communications vol 31 no 11 pp 2676ndash2684 2008
[13] Open IPTV Forum Functional Architecture [V2 1] [2011-03-15] httpwwwoipftvspecificationshtml
[14] httpwwwmclhusimjursonovics
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
International Journal of Digital Multimedia Broadcasting 3
Channel119885119899
119885119899
minus1
119861 lowast119861119899
119889
119889 119889
119862119899
++119878119899
119878119899
TX RX119877119899 119877
998400119899
Figure 3 Channel model
minimizing packet losses First we describe the channel andthe bandwidth models
41The ChannelModel Considering the requirements aboveand the attributes of aWLAN transport we decided to intro-duce a discrete-time channel and a Markov model for themathematical description of the UNIT-17 interface (AV datastreams in the Access and residential Networks)We considernot only the packet arrival but also the retransmission trafficas a discrete-time stochastic process and we prove that it is ahomogeneous Markov chain
Let 119878119899
= 0 for all 119899 represent the number of sentpackets at the time 119899 on Figure 3 TX the multicast contentdelivery function119885
119899
= 0 1 for all 119899 an additive white noisein the wireless communication channel d the transmissiondelay and RX the open IPTV terminal function (OITF) Thereceived packets are expressed with 119877
119899
= 0 for successfulpacket transmission 1 for packet loss We assume that thereceiver detects a packet loss (by checking the sequencenumbers of packets) and requests the retransmission119861
119899
= 1for a packet retransmission 0 otherwise of every lost packetsonly once from the B Fast Channel ChangeRetransmissionserver We also assume that the retransmission requestcommunication is protected by an error-free protocol likeTCP By this definition we obtain
119861119899
= 119877119899
= 119885119899minus119889
(1)
The 119861119899
signal travels through the same wireless channeltherefore it is also effected by the same 119885
119899
noise and it maybe also lost We model this effect by expressing the receivedcorrection signal119862
119899
= 0 1with three operators an inverter(minus1) a multiplier (lowast) and a channel transmission delay (d)These functions enable us to assign the value of 1 for 119862
119899
onlyin the case when the retransmission signal is not effected bythe channel noise (not lost) and the value of 0 otherwise
The final received and corrected signal 1198771015840119899
= 0 for suc-cessful packet transmission 1 for packet loss (unsuccessfulretransmission) and 2 for successful packet retransmissionis
1198771015840
119899
= 119877119899minus119889
+ 119862119899
= 119878119899minus2119889
+ 119885119899minus2119889
+ 119861119899minus119889
lowast 119885minus1
119899minus119889
= 119878119899minus2119889
+ 119885119899minus2119889
+ 119885119899minus2119889
lowast 119885minus1
119899minus119889
(2)
Let us observe that the first term of the addition equalsto 0 by definition and the last term equals to the samplingof the 119885
119899
white noise with its own delayed signal Theautocorrelation of the white noise is zero for all nonzero timeshifts [7] therefore 1198771015840
119899
can be described as a sequence ofan independent random variable which satisfies the Markovproperty119883
119899
def= 1198771015840
119899
is a homogeneous Markov chainLet119883119899
= 0 if the 119899th packet is received correctly119883119899
= 1 ifthe 119899th packet is lost and has not been retransmitted finally119883119899
= 2 if the 119899th packet is successfully retransmitted afterloss We analyzed and described a three-state Markov modelin our previous publication therefore we simply list mostimportant properties For a detailed discussion including theresolution of (3)ndash(6) and for themeaning of the probabilitiesplease refer to our former paper [8]
The transition matrix
[
[
1 minus 11990101
minus 11990102
11990101
11990102
11990110
1 minus 11990110
minus 11990112
11990112
11990120
11990121
1 minus 11990120
minus 11990121
]
]
(3)
The steady-state packet loss rate
119875LOSSsteady = (11990102
11990121
+ 11990101
11990120
+ 11990101
11990121
)
times (11990101
11990120
+ 11990101
11990121
+ 11990102
11990121
+ 11990110
11990102
+ 11990112
11990101
+11990112
11990102
+ 11990121
11990110
+ 11990120
11990110
+ 11990120
11990112
)minus1
(4)
The steady-state packet retransmission rate
119875RETsteady = (11990110
11990102
+ 11990112
11990101
+ 11990112
11990102
)
times (11990101
11990120
+ 11990101
11990121
+ 11990102
11990121
+ 11990110
11990102
+ 11990112
11990101
+11990112
11990102
+ 11990121
11990110
+ 11990120
11990110
+ 11990120
11990112
)minus1
(5)
And the probability of loss burst with length 119897
119875RETburst (119897) = (1 minus 11990120
minus 11990121
)119897minus1
(11990120
+ 11990121
) (6)
42TheModel of Bandwidth Limitation Theprevious sectionintroduced how the wireless channel affects the packet trans-mission and now we are going to analyze the the bandwidthallocation in IPTV solutions
We introduce three planes of the IPTV packet trans-mission The transmitter plane represents the providerrsquosnetwork receiver plane represents the OITF and playout planrepresents the content presentation within the OITF Theevents of the packet loss usually show a burstiness in wirelesscommunication [9ndash11] Therefore we investigate an intra-burst loss on Figure 4 after the first 119896 consecutive packetlosses (b) the receiver requests them for retransmission (c)which packets are delivered within the allocated bandwidthfor packet transmission (d) to the presentation device (e) Fora successful retransmission all retransmitted packet should
4 International Journal of Digital Multimedia Broadcasting
Bandwidth
(c) Retransmitted
(a) Sent packets
(d) Ret received
(b) Lost packets
(e) Played out
Transmitter
Receiver
Playout
1 2
1 2
RET
AVdata
AVdata
AVdata
1
middot middot middot
middot middot middot
119896
1 2 middot middot middot 119896
119896
middot middot middot 119896
RET 1 2 middot middot middot 119896
119905
119905
119905119879ret119863ret 119879playo119863playo
Figure 4 Intraburst retransmission
arrive earlier than their presentation time (119879ret le 119879playo)Expressing this condition with the durations
119863pkg + RTT + 119896119861AV119861RET
119863pkg lt (119896 minus 1)119863pkg + 119863playo (7)
where 119863pkg is the average transmission time for a packet119877119879119879 is the round-trip time 119861AV and 119861RET are the allocatedbandwidths on the communication channel and 119863playo isthe packet playout buffer in the OITF The formula ofthe maximal number of consecutive packets which can besuccessfully retransmitted is defined as
119896RETintramax (119861RET119861AV
) =
119863playo minus 2119863pkg minus RTT119863pkg ((1 (119861RET119861AV)) minus 1)
(8)
The actual throughput of the AV stream may vary byinstallations therefore we expressed this value as a ratio of119861RET and 119861AV
Second we highlight the barrier of interburst behavioron Figure 5 After a loss of long burst the retransmissionbandwidth is occupied by the traffic of the retransmittedpackets even if there is no other packet loss at the time inthe video stream This means that a loss event blocks theretransmission channel We are interested in the followingquestion assuming a 119896 lt 119896RETintramax long burst of loss afterhowmany packets (119899) can a new loss burst occurwhichwouldbe also successfully retransmitted (eg what is the minimaldistance (119899 minus 119896) between two loss bursts if the first burst lastsfor 119896 packets) Now 119879ret119896
def= 119879playo119896 and 119879ret119899 le 119879playo119899
Figure 5 shows that
119863pkg + RTT + (119896 + 1)119861AV119861RET
119863pkg lt (119899 minus 1)119863pkg + 119863playo
(9)
where 119899 gt 119896 Expressing 119899
119899RETintermin (119896119861AV119861RET
) = (119896 + 1)119861AV119861RET
+
RTT minus 119863playo
119863pkg+ 2
(10)
119879playo 119899
119879ret119899119879ret119896119863ret119896
Bandwidth
(c) Retransmitted
(a) Sent packets
(d) Ret received
(b) Lost packets (e) Played out
Transmitter
Receiver
Playout
RET
AVdata
AVdata
AVdata
middot middot middot
middot middot middot middot middot middot middot middot middot middot middot middot
middot middot middot middot middot middot
middot middot middot middot middot middot middot middot middot middot middot middot
middot middot middot
1 119896
RET 11
middot middot middot
119896119896
119905
119905
119905
119863playo
119899
1 middot middot middot 119896 119899
119899
119899
1 119896 119899
middot middot middot middot middot middot middot middot middot
middot middot middotmiddot middot middotmiddot middot middot
middot middot middot
Figure 5 Interburst retransmission
0
20
40
60
80
100
120
0 005 01 015 02 025 03 035 04
Num
ber o
f pac
kets
119861RET 119861AV
119899RETintermin 119863playo = 1000ms119896RETintramax 119863playo = 1000ms119899RETintermin 119863playo = 800ms119896RETintramax 119863playo = 800ms119899RETintermin 119863playo = 600ms119896RETintramax 119863playo = 600ms119899RETintermin 119863playo = 400ms119896RETintramax 119863playo = 400ms119899RETintermin 119863playo = 200ms119896RETintramax 119863playo = 200ms
Figure 6 Intra- and interburst limitation
We present the above declared functions (8)ndash(10) onFigure 6 This graph shows that at small values of 119861AV119861RET(0ndash15) the effect of the inter-burst blocking is greaterFor example at 5 the maximal consecutive burst lengthis 10 packets and the retransmission channel is blocked bythis traffic for 30 packets Above 20 this effect becomesinsignificant
We also state that with grater playout buffer (119863playo) RETis able to correct larger loss bursts at the price of a greater end-to-end delay
International Journal of Digital Multimedia Broadcasting 5
5 The Optimal Retransmission
In this sectionwe introduce a newmethod for retransmissionbandwidth allocation based on our models with the aimof achieving a better video quality The method is realizedin a test environment and our hypothesis is measured andproved
Traditional RET algorithms request all lost packets there-fore they have to implement a network layer traffic shapingto fit the actual retransmission throughput into the allocatedbandwidth This is usually done by packet queuing whichincreases the overall packet retransmission time thereforethe probability of a retransmission packet arrives late (afterits playout time) is great Several papers addressed thisproblem [12] introducing a selective retransmission protocolby evaluating the traffic on the application level and assigningpriority and importance for each packet retransmissions
The main advantages of our method are the minimaladditional delay the low resource needs and the consid-eration of the wireless channel Our method assess theRET mechanism on the network layer skips (forbids) theretransmission requests of a lost packet according to theabove described intra- and interburst channel blocking andtakes the special properties of the wireless channel intoconsideration Our algorithm consists of three steps
(1) The packet arrival process is continuously monitoredfor packet loss
(2) 119896RETintramax and 119899RETintermin are calculated(3) A lost packet is requested retransmission only if the
intra- and inter-burst channel blocking do not for-bid the retransmission otherwise the retransmissionrequest is skipped
51 Empirical Evaluation To evaluate our model and meth-ods we followed theOIPF system architecture [13] and imple-mented the following OIPF functions in a test environment(source codes are available on [14])
(1) Multicast content delivery function ser c++ applica-tion generates UDPRTP multicast traffic and imple-ments a simple control protocol hosted by an x86Linux server connected to the core network ofDeutsche Telekom (DT)
(2) RET server ret c++ application stores the multicasttraffic in a circular buffer and implements a simpleretransmission request protocol hosted by the samex86 Linux server
(3) Unit-17 interface part a was realized by the ADSL2+access network of DT and was provided by a DLinkADSL modem
(4) Unit-17 interface part b was realized by a 80211 bWLAN network and was provided by a Cisco 1200series wireless access point (AP) connected to theADSL modem
(5) OITF cli c++ application implements a simple mul-ticast receiver and controls functions of the multicast
0
50
100
150
200
0 50 100 150 200
Thro
ughp
ut (k
bps)
Time (s)Reserved bandwidthRET allRET skip
Figure 7 Retransmission throughput
content delivery function and RET server Hosted ona x86 Linux laptop connected to the Cisco AP
Figure 7 shows the throughput of the retransmissionstream in two cases traditional retransmission (RET all alllost packets were requested for retransmission) and our newretransmission method (RET skip retransmission requestsmay be skipped based on the actual parameters of thechannel) It can be clearly seen that our algorithm kept theretransmission throughput under its dedicated bandwidthwhich ensured the in-time delivery of the retransmissionpackets however we intentionally skipped those retransmis-sion requests which in time delivery would not be ensureddue to the channel blocking (intra- and inter-burst effect)
Themain benefit of ourmethod is showed by Figure 8Wecompared the total packet loss rate in the above mentionedtwo cases and we found that with the smart skipping ofpacket retransmission requests we were able to achievea better (smaller) loss rate then retransmitting all of thepacketsOurmethod avoided the effect of late retransmissionIF a packet is requested for retransmission without ensuringthe necessary transport bandwidth then it may delay furtherretransmission requests which may arrive to late after theirplayout time This causes an inefficient retransmission band-width utilization which increases the overall packet loss rate(on the playout plane)
52 The Effect of the Intraburst Limitation Let us analyzeour results theoretically as well In this and in the upcomingsection we characterize the Unit-17 interface with the transi-tion matrix of our Markov model and the design attributesof the access network Applying the intra- and inter-burstlimitations on our model we derive the probability of theretransmission skip caused by our algorithm Finally weexpress the overall packet loss rate which is a key indicatorfor the quality of the video transmission
6 International Journal of Digital Multimedia Broadcasting
0
001
002
003
004
005
006
007
008
009
01
0 50 100 150 200
Ove
rall
pack
et lo
ss ra
te
Time (s)
RET allRET skip
Figure 8 Overall packet loss rate
The intra-burst limitations have a significant short-timeeffect if the distance of the burst losses is great (119901
00
rarr
1) Our question is about the probability of retransmissionpacket skip First we calculate the maximal number ofconsecutive retransmission requests
119896RETintramax =
119863playo minus 2119863pkg minus RTT119863pkg ((119861AV119861RET) minus 1)
(11)
The probability of 119897 long retransmission request is givenby the Markov model (6) We calculate the probability of119898 packet skips if the retransmission burst is greater than119896119877burstmax which is
119875skip (119898) = (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1+119898
(12)
The overall probability of a packet skip is given by
119875skipintra = lim119894rarrinfin
infin
sum
119894=1
119875skip (119894)
119894
= lim119894rarrinfin
infin
sum
119894=1
(11990120
+ 11990121
) (1minus 11990120
minus 11987521
)119896RETintramaxminus1+119894
119894
= (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1
times lim119894rarrinfin
infin
sum
119894=1
(1 minus 11990120
minus 11990121
)119894
119894
(13)
Let us observe that the last sum can be expressed as aspecial form of the polylogarithm (also known as Jonquierersquosfunction)
119871119894119904
(119911) |119904=1
=
infin
sum
119896=1
119911119896
119896119904|119904=1
=
infin
sum
119896=1
119911119896
119896 (14)
for every minus1 le 119911 lt 1 (1 minus 11990120
minus 11987521
satisfies this criteriaTherefore using the well-known formula of 119871119894
1
(119911) = minus ln(1 minus
119911) the equation can be expressed in a closed form
119875skipintra = (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1
times (minus1) ln (1 minus (1 minus 11990120
minus 11990121
))
= (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1
times (minus1) ln (11990120
+ 11990121
)
(15)
The overall packet loss can be expressed as a sum of theprobability of packet skip (15) and the steady state probabilityof packet loss (4)
53 The Effect of the Interburst Limitation We ask the samequestion as in the previous section what is the probability ofpacket skip Let us assume that the first burst is small enoughto be retransmitted (119896 lt 119896RETintramax) The probability ofretransmission burst is of 119897 size is given by ourMarkov model(6)The probability of 119896 retransmission burst followed by 119899minus119896good transmission burst and a second retransmission
119891 (119896 119899) = (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896minus1
11990120
times (11990101
+ 11990102
)
times (1 minus 11990101
minus 11990102
)119899minus119896minus1
11990102
(16)
The first packet of the second retransmission burst isskipped if 119899 lt 119899RETintermin(119896) From this we can calculatethe probability of one packet skip for 119896
119875skipinterk =
119899RETintermin(119896)
sum
119894=119896
119891 (119896 119894) (17)
For the overal packet skip probability we have to summa-rize (17) for every 119896 lt 119896Rburstmax
119875skipinter =
119896RETintramax
sum
119895=1
119899RETintermin(119895)
sum
119894=119895
119891 (119895 119894) (18)
The overall packet loss can be expressed as a sum of theprobability of packet skip (18) and the steady-state probabilityof packet loss (4)
6 Conclusion
In this paper we highlighted the relevance of the wirelesstransport in todayrsquos IPTV solutions and we pointed out thatitrsquos and the access networkrsquos combined effect on the band-width management is not discussed deeply by publicationsOur general research project targets this specific area byintroducing several optimization methods from which wepresented one the optimization of the packet retransmissionon the previous pages
We created a newdiscrete-time channelmodel to describethe effect of the burst losses on the IPTV service quality and
International Journal of Digital Multimedia Broadcasting 7
the role of the packet retransmission We proved that it isa Markov Chain and as part of our results we expressedand evaluated its most important quantitative parametersUsing our model we also introduced an algorithm forretransmission optimization in IPTV solutions over WLANhome networks
As a further evaluation of our results we created atestbed in alignment with the OPIF system architecture andwe performed the empirical analysis of our channel modeland methods We showed that our concept improved theoverall packet loss characteristics Furthermore we encloseda mathematical analysis of our algorithm and we derived thetheoretical packet loss probabilities to support our measure-ments
In the next research phases we are going to investigateintroduce and leverage our theoretical results of throughputmanagement in the adaptive bitrate streaming technologiesfor IPTV solutions and we are going to evaluate our channelmodel in the media access control layer of the wireless accesstechnologies
References
[1] Open IPTV Forum ldquoRelease 2 Specificationsrdquo httpwwwoipftvspecificationshtml
[2] A Abramowski ldquoTowards H 265 video coding standardrdquo inProceedings of the Photonics Applications in Astronomy Commu-nications Industry and High-Energy Physics Experiments vol8008 of Proceedings of the SPIE 2011
[3] H Benot Digital Television Satellite Cable Terrestrial IptvMobile Tv in the Dvb Framework Focal Press 2008
[4] MMHassani S V Jalali andAAkbari ldquoEvaluatingTCPflowsbehavior by a mathematical model in WLANrdquo Journal of BasicandApplied Scientific Research vol 2 no 3 pp 2809ndash2814 2012
[5] A C Begen ldquoError control for IPTV over xDSL networksrdquoin Proceedings of the 5th IEEE Consumer Communications andNetworking Conference (CCNC rsquo08) pp 632ndash637 January 2008
[6] H Bobarshad M Van Der Schaar and M R Shikh-BahaeildquoA low-complexity analytical modeling for cross-layer adaptiveerror protection in video over WLANrdquo IEEE Transactions onMultimedia vol 12 no 5 pp 427ndash438 2010
[7] A Papoulis Probability Random Variables and Stohastic Pro-cesses McGraw-Hill 1965
[8] T Jursonovics and S Imre ldquoAnalysis of a newMarkovmodel forpacket loss characterization in IPTV Solutionsrdquo Infocommuni-cations Journal vol 2 pp 28ndash33 2011
[9] T Jursonovics and Zs Butyka ldquoThe implementation and analy-sis of interactive multimedia mobile servicesrdquo in Proceedings ofthe 11th Microcoll Conference pp 149ndash152 Budapest Hungary2003
[10] T Jursonovics Zs Butyka and S Imre ldquoMultimedia transmis-sion over mobile networksrdquo in International Conference on Soft-ware Telecommunications and Computer Networks (SoftCOMrsquo04) pp 453ndash457 Dubrovnik Croatia Ancona Italy 2004
[11] A Nafaa T Taleb and L Murphy ldquoForward error correctionstrategies for media streaming over wireless networksrdquo IEEECommunications Magazine vol 46 no 1 pp 72ndash79 2008
[12] A Huszak and S Imre ldquoSource controlled semi-reliable mul-timedia streaming using selective retransmission in DCCPIP
networksrdquo Computer Communications vol 31 no 11 pp 2676ndash2684 2008
[13] Open IPTV Forum Functional Architecture [V2 1] [2011-03-15] httpwwwoipftvspecificationshtml
[14] httpwwwmclhusimjursonovics
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
4 International Journal of Digital Multimedia Broadcasting
Bandwidth
(c) Retransmitted
(a) Sent packets
(d) Ret received
(b) Lost packets
(e) Played out
Transmitter
Receiver
Playout
1 2
1 2
RET
AVdata
AVdata
AVdata
1
middot middot middot
middot middot middot
119896
1 2 middot middot middot 119896
119896
middot middot middot 119896
RET 1 2 middot middot middot 119896
119905
119905
119905119879ret119863ret 119879playo119863playo
Figure 4 Intraburst retransmission
arrive earlier than their presentation time (119879ret le 119879playo)Expressing this condition with the durations
119863pkg + RTT + 119896119861AV119861RET
119863pkg lt (119896 minus 1)119863pkg + 119863playo (7)
where 119863pkg is the average transmission time for a packet119877119879119879 is the round-trip time 119861AV and 119861RET are the allocatedbandwidths on the communication channel and 119863playo isthe packet playout buffer in the OITF The formula ofthe maximal number of consecutive packets which can besuccessfully retransmitted is defined as
119896RETintramax (119861RET119861AV
) =
119863playo minus 2119863pkg minus RTT119863pkg ((1 (119861RET119861AV)) minus 1)
(8)
The actual throughput of the AV stream may vary byinstallations therefore we expressed this value as a ratio of119861RET and 119861AV
Second we highlight the barrier of interburst behavioron Figure 5 After a loss of long burst the retransmissionbandwidth is occupied by the traffic of the retransmittedpackets even if there is no other packet loss at the time inthe video stream This means that a loss event blocks theretransmission channel We are interested in the followingquestion assuming a 119896 lt 119896RETintramax long burst of loss afterhowmany packets (119899) can a new loss burst occurwhichwouldbe also successfully retransmitted (eg what is the minimaldistance (119899 minus 119896) between two loss bursts if the first burst lastsfor 119896 packets) Now 119879ret119896
def= 119879playo119896 and 119879ret119899 le 119879playo119899
Figure 5 shows that
119863pkg + RTT + (119896 + 1)119861AV119861RET
119863pkg lt (119899 minus 1)119863pkg + 119863playo
(9)
where 119899 gt 119896 Expressing 119899
119899RETintermin (119896119861AV119861RET
) = (119896 + 1)119861AV119861RET
+
RTT minus 119863playo
119863pkg+ 2
(10)
119879playo 119899
119879ret119899119879ret119896119863ret119896
Bandwidth
(c) Retransmitted
(a) Sent packets
(d) Ret received
(b) Lost packets (e) Played out
Transmitter
Receiver
Playout
RET
AVdata
AVdata
AVdata
middot middot middot
middot middot middot middot middot middot middot middot middot middot middot middot
middot middot middot middot middot middot
middot middot middot middot middot middot middot middot middot middot middot middot
middot middot middot
1 119896
RET 11
middot middot middot
119896119896
119905
119905
119905
119863playo
119899
1 middot middot middot 119896 119899
119899
119899
1 119896 119899
middot middot middot middot middot middot middot middot middot
middot middot middotmiddot middot middotmiddot middot middot
middot middot middot
Figure 5 Interburst retransmission
0
20
40
60
80
100
120
0 005 01 015 02 025 03 035 04
Num
ber o
f pac
kets
119861RET 119861AV
119899RETintermin 119863playo = 1000ms119896RETintramax 119863playo = 1000ms119899RETintermin 119863playo = 800ms119896RETintramax 119863playo = 800ms119899RETintermin 119863playo = 600ms119896RETintramax 119863playo = 600ms119899RETintermin 119863playo = 400ms119896RETintramax 119863playo = 400ms119899RETintermin 119863playo = 200ms119896RETintramax 119863playo = 200ms
Figure 6 Intra- and interburst limitation
We present the above declared functions (8)ndash(10) onFigure 6 This graph shows that at small values of 119861AV119861RET(0ndash15) the effect of the inter-burst blocking is greaterFor example at 5 the maximal consecutive burst lengthis 10 packets and the retransmission channel is blocked bythis traffic for 30 packets Above 20 this effect becomesinsignificant
We also state that with grater playout buffer (119863playo) RETis able to correct larger loss bursts at the price of a greater end-to-end delay
International Journal of Digital Multimedia Broadcasting 5
5 The Optimal Retransmission
In this sectionwe introduce a newmethod for retransmissionbandwidth allocation based on our models with the aimof achieving a better video quality The method is realizedin a test environment and our hypothesis is measured andproved
Traditional RET algorithms request all lost packets there-fore they have to implement a network layer traffic shapingto fit the actual retransmission throughput into the allocatedbandwidth This is usually done by packet queuing whichincreases the overall packet retransmission time thereforethe probability of a retransmission packet arrives late (afterits playout time) is great Several papers addressed thisproblem [12] introducing a selective retransmission protocolby evaluating the traffic on the application level and assigningpriority and importance for each packet retransmissions
The main advantages of our method are the minimaladditional delay the low resource needs and the consid-eration of the wireless channel Our method assess theRET mechanism on the network layer skips (forbids) theretransmission requests of a lost packet according to theabove described intra- and interburst channel blocking andtakes the special properties of the wireless channel intoconsideration Our algorithm consists of three steps
(1) The packet arrival process is continuously monitoredfor packet loss
(2) 119896RETintramax and 119899RETintermin are calculated(3) A lost packet is requested retransmission only if the
intra- and inter-burst channel blocking do not for-bid the retransmission otherwise the retransmissionrequest is skipped
51 Empirical Evaluation To evaluate our model and meth-ods we followed theOIPF system architecture [13] and imple-mented the following OIPF functions in a test environment(source codes are available on [14])
(1) Multicast content delivery function ser c++ applica-tion generates UDPRTP multicast traffic and imple-ments a simple control protocol hosted by an x86Linux server connected to the core network ofDeutsche Telekom (DT)
(2) RET server ret c++ application stores the multicasttraffic in a circular buffer and implements a simpleretransmission request protocol hosted by the samex86 Linux server
(3) Unit-17 interface part a was realized by the ADSL2+access network of DT and was provided by a DLinkADSL modem
(4) Unit-17 interface part b was realized by a 80211 bWLAN network and was provided by a Cisco 1200series wireless access point (AP) connected to theADSL modem
(5) OITF cli c++ application implements a simple mul-ticast receiver and controls functions of the multicast
0
50
100
150
200
0 50 100 150 200
Thro
ughp
ut (k
bps)
Time (s)Reserved bandwidthRET allRET skip
Figure 7 Retransmission throughput
content delivery function and RET server Hosted ona x86 Linux laptop connected to the Cisco AP
Figure 7 shows the throughput of the retransmissionstream in two cases traditional retransmission (RET all alllost packets were requested for retransmission) and our newretransmission method (RET skip retransmission requestsmay be skipped based on the actual parameters of thechannel) It can be clearly seen that our algorithm kept theretransmission throughput under its dedicated bandwidthwhich ensured the in-time delivery of the retransmissionpackets however we intentionally skipped those retransmis-sion requests which in time delivery would not be ensureddue to the channel blocking (intra- and inter-burst effect)
Themain benefit of ourmethod is showed by Figure 8Wecompared the total packet loss rate in the above mentionedtwo cases and we found that with the smart skipping ofpacket retransmission requests we were able to achievea better (smaller) loss rate then retransmitting all of thepacketsOurmethod avoided the effect of late retransmissionIF a packet is requested for retransmission without ensuringthe necessary transport bandwidth then it may delay furtherretransmission requests which may arrive to late after theirplayout time This causes an inefficient retransmission band-width utilization which increases the overall packet loss rate(on the playout plane)
52 The Effect of the Intraburst Limitation Let us analyzeour results theoretically as well In this and in the upcomingsection we characterize the Unit-17 interface with the transi-tion matrix of our Markov model and the design attributesof the access network Applying the intra- and inter-burstlimitations on our model we derive the probability of theretransmission skip caused by our algorithm Finally weexpress the overall packet loss rate which is a key indicatorfor the quality of the video transmission
6 International Journal of Digital Multimedia Broadcasting
0
001
002
003
004
005
006
007
008
009
01
0 50 100 150 200
Ove
rall
pack
et lo
ss ra
te
Time (s)
RET allRET skip
Figure 8 Overall packet loss rate
The intra-burst limitations have a significant short-timeeffect if the distance of the burst losses is great (119901
00
rarr
1) Our question is about the probability of retransmissionpacket skip First we calculate the maximal number ofconsecutive retransmission requests
119896RETintramax =
119863playo minus 2119863pkg minus RTT119863pkg ((119861AV119861RET) minus 1)
(11)
The probability of 119897 long retransmission request is givenby the Markov model (6) We calculate the probability of119898 packet skips if the retransmission burst is greater than119896119877burstmax which is
119875skip (119898) = (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1+119898
(12)
The overall probability of a packet skip is given by
119875skipintra = lim119894rarrinfin
infin
sum
119894=1
119875skip (119894)
119894
= lim119894rarrinfin
infin
sum
119894=1
(11990120
+ 11990121
) (1minus 11990120
minus 11987521
)119896RETintramaxminus1+119894
119894
= (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1
times lim119894rarrinfin
infin
sum
119894=1
(1 minus 11990120
minus 11990121
)119894
119894
(13)
Let us observe that the last sum can be expressed as aspecial form of the polylogarithm (also known as Jonquierersquosfunction)
119871119894119904
(119911) |119904=1
=
infin
sum
119896=1
119911119896
119896119904|119904=1
=
infin
sum
119896=1
119911119896
119896 (14)
for every minus1 le 119911 lt 1 (1 minus 11990120
minus 11987521
satisfies this criteriaTherefore using the well-known formula of 119871119894
1
(119911) = minus ln(1 minus
119911) the equation can be expressed in a closed form
119875skipintra = (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1
times (minus1) ln (1 minus (1 minus 11990120
minus 11990121
))
= (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1
times (minus1) ln (11990120
+ 11990121
)
(15)
The overall packet loss can be expressed as a sum of theprobability of packet skip (15) and the steady state probabilityof packet loss (4)
53 The Effect of the Interburst Limitation We ask the samequestion as in the previous section what is the probability ofpacket skip Let us assume that the first burst is small enoughto be retransmitted (119896 lt 119896RETintramax) The probability ofretransmission burst is of 119897 size is given by ourMarkov model(6)The probability of 119896 retransmission burst followed by 119899minus119896good transmission burst and a second retransmission
119891 (119896 119899) = (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896minus1
11990120
times (11990101
+ 11990102
)
times (1 minus 11990101
minus 11990102
)119899minus119896minus1
11990102
(16)
The first packet of the second retransmission burst isskipped if 119899 lt 119899RETintermin(119896) From this we can calculatethe probability of one packet skip for 119896
119875skipinterk =
119899RETintermin(119896)
sum
119894=119896
119891 (119896 119894) (17)
For the overal packet skip probability we have to summa-rize (17) for every 119896 lt 119896Rburstmax
119875skipinter =
119896RETintramax
sum
119895=1
119899RETintermin(119895)
sum
119894=119895
119891 (119895 119894) (18)
The overall packet loss can be expressed as a sum of theprobability of packet skip (18) and the steady-state probabilityof packet loss (4)
6 Conclusion
In this paper we highlighted the relevance of the wirelesstransport in todayrsquos IPTV solutions and we pointed out thatitrsquos and the access networkrsquos combined effect on the band-width management is not discussed deeply by publicationsOur general research project targets this specific area byintroducing several optimization methods from which wepresented one the optimization of the packet retransmissionon the previous pages
We created a newdiscrete-time channelmodel to describethe effect of the burst losses on the IPTV service quality and
International Journal of Digital Multimedia Broadcasting 7
the role of the packet retransmission We proved that it isa Markov Chain and as part of our results we expressedand evaluated its most important quantitative parametersUsing our model we also introduced an algorithm forretransmission optimization in IPTV solutions over WLANhome networks
As a further evaluation of our results we created atestbed in alignment with the OPIF system architecture andwe performed the empirical analysis of our channel modeland methods We showed that our concept improved theoverall packet loss characteristics Furthermore we encloseda mathematical analysis of our algorithm and we derived thetheoretical packet loss probabilities to support our measure-ments
In the next research phases we are going to investigateintroduce and leverage our theoretical results of throughputmanagement in the adaptive bitrate streaming technologiesfor IPTV solutions and we are going to evaluate our channelmodel in the media access control layer of the wireless accesstechnologies
References
[1] Open IPTV Forum ldquoRelease 2 Specificationsrdquo httpwwwoipftvspecificationshtml
[2] A Abramowski ldquoTowards H 265 video coding standardrdquo inProceedings of the Photonics Applications in Astronomy Commu-nications Industry and High-Energy Physics Experiments vol8008 of Proceedings of the SPIE 2011
[3] H Benot Digital Television Satellite Cable Terrestrial IptvMobile Tv in the Dvb Framework Focal Press 2008
[4] MMHassani S V Jalali andAAkbari ldquoEvaluatingTCPflowsbehavior by a mathematical model in WLANrdquo Journal of BasicandApplied Scientific Research vol 2 no 3 pp 2809ndash2814 2012
[5] A C Begen ldquoError control for IPTV over xDSL networksrdquoin Proceedings of the 5th IEEE Consumer Communications andNetworking Conference (CCNC rsquo08) pp 632ndash637 January 2008
[6] H Bobarshad M Van Der Schaar and M R Shikh-BahaeildquoA low-complexity analytical modeling for cross-layer adaptiveerror protection in video over WLANrdquo IEEE Transactions onMultimedia vol 12 no 5 pp 427ndash438 2010
[7] A Papoulis Probability Random Variables and Stohastic Pro-cesses McGraw-Hill 1965
[8] T Jursonovics and S Imre ldquoAnalysis of a newMarkovmodel forpacket loss characterization in IPTV Solutionsrdquo Infocommuni-cations Journal vol 2 pp 28ndash33 2011
[9] T Jursonovics and Zs Butyka ldquoThe implementation and analy-sis of interactive multimedia mobile servicesrdquo in Proceedings ofthe 11th Microcoll Conference pp 149ndash152 Budapest Hungary2003
[10] T Jursonovics Zs Butyka and S Imre ldquoMultimedia transmis-sion over mobile networksrdquo in International Conference on Soft-ware Telecommunications and Computer Networks (SoftCOMrsquo04) pp 453ndash457 Dubrovnik Croatia Ancona Italy 2004
[11] A Nafaa T Taleb and L Murphy ldquoForward error correctionstrategies for media streaming over wireless networksrdquo IEEECommunications Magazine vol 46 no 1 pp 72ndash79 2008
[12] A Huszak and S Imre ldquoSource controlled semi-reliable mul-timedia streaming using selective retransmission in DCCPIP
networksrdquo Computer Communications vol 31 no 11 pp 2676ndash2684 2008
[13] Open IPTV Forum Functional Architecture [V2 1] [2011-03-15] httpwwwoipftvspecificationshtml
[14] httpwwwmclhusimjursonovics
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
International Journal of Digital Multimedia Broadcasting 5
5 The Optimal Retransmission
In this sectionwe introduce a newmethod for retransmissionbandwidth allocation based on our models with the aimof achieving a better video quality The method is realizedin a test environment and our hypothesis is measured andproved
Traditional RET algorithms request all lost packets there-fore they have to implement a network layer traffic shapingto fit the actual retransmission throughput into the allocatedbandwidth This is usually done by packet queuing whichincreases the overall packet retransmission time thereforethe probability of a retransmission packet arrives late (afterits playout time) is great Several papers addressed thisproblem [12] introducing a selective retransmission protocolby evaluating the traffic on the application level and assigningpriority and importance for each packet retransmissions
The main advantages of our method are the minimaladditional delay the low resource needs and the consid-eration of the wireless channel Our method assess theRET mechanism on the network layer skips (forbids) theretransmission requests of a lost packet according to theabove described intra- and interburst channel blocking andtakes the special properties of the wireless channel intoconsideration Our algorithm consists of three steps
(1) The packet arrival process is continuously monitoredfor packet loss
(2) 119896RETintramax and 119899RETintermin are calculated(3) A lost packet is requested retransmission only if the
intra- and inter-burst channel blocking do not for-bid the retransmission otherwise the retransmissionrequest is skipped
51 Empirical Evaluation To evaluate our model and meth-ods we followed theOIPF system architecture [13] and imple-mented the following OIPF functions in a test environment(source codes are available on [14])
(1) Multicast content delivery function ser c++ applica-tion generates UDPRTP multicast traffic and imple-ments a simple control protocol hosted by an x86Linux server connected to the core network ofDeutsche Telekom (DT)
(2) RET server ret c++ application stores the multicasttraffic in a circular buffer and implements a simpleretransmission request protocol hosted by the samex86 Linux server
(3) Unit-17 interface part a was realized by the ADSL2+access network of DT and was provided by a DLinkADSL modem
(4) Unit-17 interface part b was realized by a 80211 bWLAN network and was provided by a Cisco 1200series wireless access point (AP) connected to theADSL modem
(5) OITF cli c++ application implements a simple mul-ticast receiver and controls functions of the multicast
0
50
100
150
200
0 50 100 150 200
Thro
ughp
ut (k
bps)
Time (s)Reserved bandwidthRET allRET skip
Figure 7 Retransmission throughput
content delivery function and RET server Hosted ona x86 Linux laptop connected to the Cisco AP
Figure 7 shows the throughput of the retransmissionstream in two cases traditional retransmission (RET all alllost packets were requested for retransmission) and our newretransmission method (RET skip retransmission requestsmay be skipped based on the actual parameters of thechannel) It can be clearly seen that our algorithm kept theretransmission throughput under its dedicated bandwidthwhich ensured the in-time delivery of the retransmissionpackets however we intentionally skipped those retransmis-sion requests which in time delivery would not be ensureddue to the channel blocking (intra- and inter-burst effect)
Themain benefit of ourmethod is showed by Figure 8Wecompared the total packet loss rate in the above mentionedtwo cases and we found that with the smart skipping ofpacket retransmission requests we were able to achievea better (smaller) loss rate then retransmitting all of thepacketsOurmethod avoided the effect of late retransmissionIF a packet is requested for retransmission without ensuringthe necessary transport bandwidth then it may delay furtherretransmission requests which may arrive to late after theirplayout time This causes an inefficient retransmission band-width utilization which increases the overall packet loss rate(on the playout plane)
52 The Effect of the Intraburst Limitation Let us analyzeour results theoretically as well In this and in the upcomingsection we characterize the Unit-17 interface with the transi-tion matrix of our Markov model and the design attributesof the access network Applying the intra- and inter-burstlimitations on our model we derive the probability of theretransmission skip caused by our algorithm Finally weexpress the overall packet loss rate which is a key indicatorfor the quality of the video transmission
6 International Journal of Digital Multimedia Broadcasting
0
001
002
003
004
005
006
007
008
009
01
0 50 100 150 200
Ove
rall
pack
et lo
ss ra
te
Time (s)
RET allRET skip
Figure 8 Overall packet loss rate
The intra-burst limitations have a significant short-timeeffect if the distance of the burst losses is great (119901
00
rarr
1) Our question is about the probability of retransmissionpacket skip First we calculate the maximal number ofconsecutive retransmission requests
119896RETintramax =
119863playo minus 2119863pkg minus RTT119863pkg ((119861AV119861RET) minus 1)
(11)
The probability of 119897 long retransmission request is givenby the Markov model (6) We calculate the probability of119898 packet skips if the retransmission burst is greater than119896119877burstmax which is
119875skip (119898) = (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1+119898
(12)
The overall probability of a packet skip is given by
119875skipintra = lim119894rarrinfin
infin
sum
119894=1
119875skip (119894)
119894
= lim119894rarrinfin
infin
sum
119894=1
(11990120
+ 11990121
) (1minus 11990120
minus 11987521
)119896RETintramaxminus1+119894
119894
= (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1
times lim119894rarrinfin
infin
sum
119894=1
(1 minus 11990120
minus 11990121
)119894
119894
(13)
Let us observe that the last sum can be expressed as aspecial form of the polylogarithm (also known as Jonquierersquosfunction)
119871119894119904
(119911) |119904=1
=
infin
sum
119896=1
119911119896
119896119904|119904=1
=
infin
sum
119896=1
119911119896
119896 (14)
for every minus1 le 119911 lt 1 (1 minus 11990120
minus 11987521
satisfies this criteriaTherefore using the well-known formula of 119871119894
1
(119911) = minus ln(1 minus
119911) the equation can be expressed in a closed form
119875skipintra = (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1
times (minus1) ln (1 minus (1 minus 11990120
minus 11990121
))
= (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1
times (minus1) ln (11990120
+ 11990121
)
(15)
The overall packet loss can be expressed as a sum of theprobability of packet skip (15) and the steady state probabilityof packet loss (4)
53 The Effect of the Interburst Limitation We ask the samequestion as in the previous section what is the probability ofpacket skip Let us assume that the first burst is small enoughto be retransmitted (119896 lt 119896RETintramax) The probability ofretransmission burst is of 119897 size is given by ourMarkov model(6)The probability of 119896 retransmission burst followed by 119899minus119896good transmission burst and a second retransmission
119891 (119896 119899) = (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896minus1
11990120
times (11990101
+ 11990102
)
times (1 minus 11990101
minus 11990102
)119899minus119896minus1
11990102
(16)
The first packet of the second retransmission burst isskipped if 119899 lt 119899RETintermin(119896) From this we can calculatethe probability of one packet skip for 119896
119875skipinterk =
119899RETintermin(119896)
sum
119894=119896
119891 (119896 119894) (17)
For the overal packet skip probability we have to summa-rize (17) for every 119896 lt 119896Rburstmax
119875skipinter =
119896RETintramax
sum
119895=1
119899RETintermin(119895)
sum
119894=119895
119891 (119895 119894) (18)
The overall packet loss can be expressed as a sum of theprobability of packet skip (18) and the steady-state probabilityof packet loss (4)
6 Conclusion
In this paper we highlighted the relevance of the wirelesstransport in todayrsquos IPTV solutions and we pointed out thatitrsquos and the access networkrsquos combined effect on the band-width management is not discussed deeply by publicationsOur general research project targets this specific area byintroducing several optimization methods from which wepresented one the optimization of the packet retransmissionon the previous pages
We created a newdiscrete-time channelmodel to describethe effect of the burst losses on the IPTV service quality and
International Journal of Digital Multimedia Broadcasting 7
the role of the packet retransmission We proved that it isa Markov Chain and as part of our results we expressedand evaluated its most important quantitative parametersUsing our model we also introduced an algorithm forretransmission optimization in IPTV solutions over WLANhome networks
As a further evaluation of our results we created atestbed in alignment with the OPIF system architecture andwe performed the empirical analysis of our channel modeland methods We showed that our concept improved theoverall packet loss characteristics Furthermore we encloseda mathematical analysis of our algorithm and we derived thetheoretical packet loss probabilities to support our measure-ments
In the next research phases we are going to investigateintroduce and leverage our theoretical results of throughputmanagement in the adaptive bitrate streaming technologiesfor IPTV solutions and we are going to evaluate our channelmodel in the media access control layer of the wireless accesstechnologies
References
[1] Open IPTV Forum ldquoRelease 2 Specificationsrdquo httpwwwoipftvspecificationshtml
[2] A Abramowski ldquoTowards H 265 video coding standardrdquo inProceedings of the Photonics Applications in Astronomy Commu-nications Industry and High-Energy Physics Experiments vol8008 of Proceedings of the SPIE 2011
[3] H Benot Digital Television Satellite Cable Terrestrial IptvMobile Tv in the Dvb Framework Focal Press 2008
[4] MMHassani S V Jalali andAAkbari ldquoEvaluatingTCPflowsbehavior by a mathematical model in WLANrdquo Journal of BasicandApplied Scientific Research vol 2 no 3 pp 2809ndash2814 2012
[5] A C Begen ldquoError control for IPTV over xDSL networksrdquoin Proceedings of the 5th IEEE Consumer Communications andNetworking Conference (CCNC rsquo08) pp 632ndash637 January 2008
[6] H Bobarshad M Van Der Schaar and M R Shikh-BahaeildquoA low-complexity analytical modeling for cross-layer adaptiveerror protection in video over WLANrdquo IEEE Transactions onMultimedia vol 12 no 5 pp 427ndash438 2010
[7] A Papoulis Probability Random Variables and Stohastic Pro-cesses McGraw-Hill 1965
[8] T Jursonovics and S Imre ldquoAnalysis of a newMarkovmodel forpacket loss characterization in IPTV Solutionsrdquo Infocommuni-cations Journal vol 2 pp 28ndash33 2011
[9] T Jursonovics and Zs Butyka ldquoThe implementation and analy-sis of interactive multimedia mobile servicesrdquo in Proceedings ofthe 11th Microcoll Conference pp 149ndash152 Budapest Hungary2003
[10] T Jursonovics Zs Butyka and S Imre ldquoMultimedia transmis-sion over mobile networksrdquo in International Conference on Soft-ware Telecommunications and Computer Networks (SoftCOMrsquo04) pp 453ndash457 Dubrovnik Croatia Ancona Italy 2004
[11] A Nafaa T Taleb and L Murphy ldquoForward error correctionstrategies for media streaming over wireless networksrdquo IEEECommunications Magazine vol 46 no 1 pp 72ndash79 2008
[12] A Huszak and S Imre ldquoSource controlled semi-reliable mul-timedia streaming using selective retransmission in DCCPIP
networksrdquo Computer Communications vol 31 no 11 pp 2676ndash2684 2008
[13] Open IPTV Forum Functional Architecture [V2 1] [2011-03-15] httpwwwoipftvspecificationshtml
[14] httpwwwmclhusimjursonovics
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
6 International Journal of Digital Multimedia Broadcasting
0
001
002
003
004
005
006
007
008
009
01
0 50 100 150 200
Ove
rall
pack
et lo
ss ra
te
Time (s)
RET allRET skip
Figure 8 Overall packet loss rate
The intra-burst limitations have a significant short-timeeffect if the distance of the burst losses is great (119901
00
rarr
1) Our question is about the probability of retransmissionpacket skip First we calculate the maximal number ofconsecutive retransmission requests
119896RETintramax =
119863playo minus 2119863pkg minus RTT119863pkg ((119861AV119861RET) minus 1)
(11)
The probability of 119897 long retransmission request is givenby the Markov model (6) We calculate the probability of119898 packet skips if the retransmission burst is greater than119896119877burstmax which is
119875skip (119898) = (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1+119898
(12)
The overall probability of a packet skip is given by
119875skipintra = lim119894rarrinfin
infin
sum
119894=1
119875skip (119894)
119894
= lim119894rarrinfin
infin
sum
119894=1
(11990120
+ 11990121
) (1minus 11990120
minus 11987521
)119896RETintramaxminus1+119894
119894
= (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1
times lim119894rarrinfin
infin
sum
119894=1
(1 minus 11990120
minus 11990121
)119894
119894
(13)
Let us observe that the last sum can be expressed as aspecial form of the polylogarithm (also known as Jonquierersquosfunction)
119871119894119904
(119911) |119904=1
=
infin
sum
119896=1
119911119896
119896119904|119904=1
=
infin
sum
119896=1
119911119896
119896 (14)
for every minus1 le 119911 lt 1 (1 minus 11990120
minus 11987521
satisfies this criteriaTherefore using the well-known formula of 119871119894
1
(119911) = minus ln(1 minus
119911) the equation can be expressed in a closed form
119875skipintra = (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1
times (minus1) ln (1 minus (1 minus 11990120
minus 11990121
))
= (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896RETintramaxminus1
times (minus1) ln (11990120
+ 11990121
)
(15)
The overall packet loss can be expressed as a sum of theprobability of packet skip (15) and the steady state probabilityof packet loss (4)
53 The Effect of the Interburst Limitation We ask the samequestion as in the previous section what is the probability ofpacket skip Let us assume that the first burst is small enoughto be retransmitted (119896 lt 119896RETintramax) The probability ofretransmission burst is of 119897 size is given by ourMarkov model(6)The probability of 119896 retransmission burst followed by 119899minus119896good transmission burst and a second retransmission
119891 (119896 119899) = (11990120
+ 11990121
) (1 minus 11990120
minus 11990121
)119896minus1
11990120
times (11990101
+ 11990102
)
times (1 minus 11990101
minus 11990102
)119899minus119896minus1
11990102
(16)
The first packet of the second retransmission burst isskipped if 119899 lt 119899RETintermin(119896) From this we can calculatethe probability of one packet skip for 119896
119875skipinterk =
119899RETintermin(119896)
sum
119894=119896
119891 (119896 119894) (17)
For the overal packet skip probability we have to summa-rize (17) for every 119896 lt 119896Rburstmax
119875skipinter =
119896RETintramax
sum
119895=1
119899RETintermin(119895)
sum
119894=119895
119891 (119895 119894) (18)
The overall packet loss can be expressed as a sum of theprobability of packet skip (18) and the steady-state probabilityof packet loss (4)
6 Conclusion
In this paper we highlighted the relevance of the wirelesstransport in todayrsquos IPTV solutions and we pointed out thatitrsquos and the access networkrsquos combined effect on the band-width management is not discussed deeply by publicationsOur general research project targets this specific area byintroducing several optimization methods from which wepresented one the optimization of the packet retransmissionon the previous pages
We created a newdiscrete-time channelmodel to describethe effect of the burst losses on the IPTV service quality and
International Journal of Digital Multimedia Broadcasting 7
the role of the packet retransmission We proved that it isa Markov Chain and as part of our results we expressedand evaluated its most important quantitative parametersUsing our model we also introduced an algorithm forretransmission optimization in IPTV solutions over WLANhome networks
As a further evaluation of our results we created atestbed in alignment with the OPIF system architecture andwe performed the empirical analysis of our channel modeland methods We showed that our concept improved theoverall packet loss characteristics Furthermore we encloseda mathematical analysis of our algorithm and we derived thetheoretical packet loss probabilities to support our measure-ments
In the next research phases we are going to investigateintroduce and leverage our theoretical results of throughputmanagement in the adaptive bitrate streaming technologiesfor IPTV solutions and we are going to evaluate our channelmodel in the media access control layer of the wireless accesstechnologies
References
[1] Open IPTV Forum ldquoRelease 2 Specificationsrdquo httpwwwoipftvspecificationshtml
[2] A Abramowski ldquoTowards H 265 video coding standardrdquo inProceedings of the Photonics Applications in Astronomy Commu-nications Industry and High-Energy Physics Experiments vol8008 of Proceedings of the SPIE 2011
[3] H Benot Digital Television Satellite Cable Terrestrial IptvMobile Tv in the Dvb Framework Focal Press 2008
[4] MMHassani S V Jalali andAAkbari ldquoEvaluatingTCPflowsbehavior by a mathematical model in WLANrdquo Journal of BasicandApplied Scientific Research vol 2 no 3 pp 2809ndash2814 2012
[5] A C Begen ldquoError control for IPTV over xDSL networksrdquoin Proceedings of the 5th IEEE Consumer Communications andNetworking Conference (CCNC rsquo08) pp 632ndash637 January 2008
[6] H Bobarshad M Van Der Schaar and M R Shikh-BahaeildquoA low-complexity analytical modeling for cross-layer adaptiveerror protection in video over WLANrdquo IEEE Transactions onMultimedia vol 12 no 5 pp 427ndash438 2010
[7] A Papoulis Probability Random Variables and Stohastic Pro-cesses McGraw-Hill 1965
[8] T Jursonovics and S Imre ldquoAnalysis of a newMarkovmodel forpacket loss characterization in IPTV Solutionsrdquo Infocommuni-cations Journal vol 2 pp 28ndash33 2011
[9] T Jursonovics and Zs Butyka ldquoThe implementation and analy-sis of interactive multimedia mobile servicesrdquo in Proceedings ofthe 11th Microcoll Conference pp 149ndash152 Budapest Hungary2003
[10] T Jursonovics Zs Butyka and S Imre ldquoMultimedia transmis-sion over mobile networksrdquo in International Conference on Soft-ware Telecommunications and Computer Networks (SoftCOMrsquo04) pp 453ndash457 Dubrovnik Croatia Ancona Italy 2004
[11] A Nafaa T Taleb and L Murphy ldquoForward error correctionstrategies for media streaming over wireless networksrdquo IEEECommunications Magazine vol 46 no 1 pp 72ndash79 2008
[12] A Huszak and S Imre ldquoSource controlled semi-reliable mul-timedia streaming using selective retransmission in DCCPIP
networksrdquo Computer Communications vol 31 no 11 pp 2676ndash2684 2008
[13] Open IPTV Forum Functional Architecture [V2 1] [2011-03-15] httpwwwoipftvspecificationshtml
[14] httpwwwmclhusimjursonovics
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
International Journal of Digital Multimedia Broadcasting 7
the role of the packet retransmission We proved that it isa Markov Chain and as part of our results we expressedand evaluated its most important quantitative parametersUsing our model we also introduced an algorithm forretransmission optimization in IPTV solutions over WLANhome networks
As a further evaluation of our results we created atestbed in alignment with the OPIF system architecture andwe performed the empirical analysis of our channel modeland methods We showed that our concept improved theoverall packet loss characteristics Furthermore we encloseda mathematical analysis of our algorithm and we derived thetheoretical packet loss probabilities to support our measure-ments
In the next research phases we are going to investigateintroduce and leverage our theoretical results of throughputmanagement in the adaptive bitrate streaming technologiesfor IPTV solutions and we are going to evaluate our channelmodel in the media access control layer of the wireless accesstechnologies
References
[1] Open IPTV Forum ldquoRelease 2 Specificationsrdquo httpwwwoipftvspecificationshtml
[2] A Abramowski ldquoTowards H 265 video coding standardrdquo inProceedings of the Photonics Applications in Astronomy Commu-nications Industry and High-Energy Physics Experiments vol8008 of Proceedings of the SPIE 2011
[3] H Benot Digital Television Satellite Cable Terrestrial IptvMobile Tv in the Dvb Framework Focal Press 2008
[4] MMHassani S V Jalali andAAkbari ldquoEvaluatingTCPflowsbehavior by a mathematical model in WLANrdquo Journal of BasicandApplied Scientific Research vol 2 no 3 pp 2809ndash2814 2012
[5] A C Begen ldquoError control for IPTV over xDSL networksrdquoin Proceedings of the 5th IEEE Consumer Communications andNetworking Conference (CCNC rsquo08) pp 632ndash637 January 2008
[6] H Bobarshad M Van Der Schaar and M R Shikh-BahaeildquoA low-complexity analytical modeling for cross-layer adaptiveerror protection in video over WLANrdquo IEEE Transactions onMultimedia vol 12 no 5 pp 427ndash438 2010
[7] A Papoulis Probability Random Variables and Stohastic Pro-cesses McGraw-Hill 1965
[8] T Jursonovics and S Imre ldquoAnalysis of a newMarkovmodel forpacket loss characterization in IPTV Solutionsrdquo Infocommuni-cations Journal vol 2 pp 28ndash33 2011
[9] T Jursonovics and Zs Butyka ldquoThe implementation and analy-sis of interactive multimedia mobile servicesrdquo in Proceedings ofthe 11th Microcoll Conference pp 149ndash152 Budapest Hungary2003
[10] T Jursonovics Zs Butyka and S Imre ldquoMultimedia transmis-sion over mobile networksrdquo in International Conference on Soft-ware Telecommunications and Computer Networks (SoftCOMrsquo04) pp 453ndash457 Dubrovnik Croatia Ancona Italy 2004
[11] A Nafaa T Taleb and L Murphy ldquoForward error correctionstrategies for media streaming over wireless networksrdquo IEEECommunications Magazine vol 46 no 1 pp 72ndash79 2008
[12] A Huszak and S Imre ldquoSource controlled semi-reliable mul-timedia streaming using selective retransmission in DCCPIP
networksrdquo Computer Communications vol 31 no 11 pp 2676ndash2684 2008
[13] Open IPTV Forum Functional Architecture [V2 1] [2011-03-15] httpwwwoipftvspecificationshtml
[14] httpwwwmclhusimjursonovics
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of