Home > Article > Multimedia Traffic over Wireless and Satellite Networks

Multimedia Traffic over Wireless and Satellite Networks

vith the continuously increasing demandble to dealthe Qosnts required by users In order to provide quality of service tontrol algorithm These mechanisms help to manage multimedia traffic guaranteeingthe system and providing Qos to theto evaluate the quality of video traffic with the mobility it is important to examining theality assessment techniques: Subjective and objective quality assessmentideo qualitygIn the last ftimedia applications are grew very fast in all networks typologiesparticular in wireless networks thattelecommunication field This trend of applications has pushed the researchers to performlot of studies in the video applications and in particular in the compression field in order tobe able of transporting this information in the network withmpact in the systerproposedder to beo transmit video traffic on different networkiveoblem in terms of bandwidth capacity and, then, aery pertoglgreat support to network serviceprovider in respecting the quality constrains otherwise nor achievabletelecommunications networks requires the integration of a lot of technologies that aim toimprove the quality of the applicatioed by theFirst oftraditional concept of best effort paradigm in the delivery of multimedia contents is notpossiblnot match with the users requirements

This typeapproach try to do its best but it is unable of guaranteeing any form of users requirementsto address thisPeblem, recently different Quality of Services architive been proposed capable of guaranteethe multimedians the users constrainsThe most famous architectures are Integrated Services and Differentiated Services thatmanage different class of sediscriminate also a cost differentiation for the customers It is easy to understand that thispe of applbasednstantly reliablption of information thatossible to have only through an appropriate network manways more quality aware and then theycordingly networks have to guarantee always more capacity in order to satisfy their usercademia, to find solutions for improving the quality of multimedia content delivered to thewell, international standards bodies, such as the international telUnion (ITU), are renewing their effort on the standardization of multimedThere are very different directions in which research has attempted to firorder to improve the quality of the rich media content deliveredMto determine efficientt is reallyimportant to know how the network behaves in terms of parameters of service for end userin order to take the correctduring the development, evaluation, construction andoperation of network services

Itimedia TrafficWireless and Satellite Networksthat is tea coding schemeused and they represent an Error Control schemen block coding Theygeneralization of the Hamming codes and they addto form code words; they also introduce127 The non-binary BCH-family codessed are the Reed Solomon(RS)codes, RS codes groups the bits intoIn order to achieve lowd for the worstchannstructures may be required to realize the powerful long codusing the Adaptive Coding(AC)[28, 29, byredundancy on the basis of channel conditions and characteristics Differeralgorithms have been proposed for video streaming applications, as the one p[28]: in that work the fading level is estimated and, then, the algorithm evaluatescoding ratio in order to actively protect packets from losses In thischannelbe immediately and efficiently improved3

2 Error perceptiAs discussed earlier, packet losses affect the quality of multimedwireless packet networks and thof quality degradation strongly vary on the basisof the meaning of the lost data For the designing of an efficient loss protection mechanism,a reliable estimation method for multimedia data is needed Providing an accurateunlined: for example, it is very useful to introduce thelock or packet conceptvalue directly related to the distortion thatbe introduced at thder by the loss ofAnalysis By Synthesis"(ABS)algorithm are illustratedthe possible distortion of each elemsteps, applied for each packetnalyzed (synthesmute the distortion caused by thed the reconstructed pictures are compared using Mean Square Error(MSE)dication of the perceptual importance of theLittle modifications of the standardsrder to imTherithmstructs theecoder operations It can be surely used for video coding, but the obtained valueson the adopted encoding (if the video willpressed with a different encoder or ifdifferent packetization policy is used, values will be very different The main disadvantageof the exposed technique consists of the interdependencies usually present between data

vidDecodingImportancevalueDecoding+concealment(synthesis)by synthesis blocks diagralated dts should beghted by its probability, and its distortion computed by the abs technique) If propagation is preshe distorintroducedin next frames should be evaluated until it is negligible(foruntil a frame isached in a mPeg stream) In thishe application of the ABS scheme is harder, due toputation of the perceptual distortionAnother scheme for distortion evalis represented by the Distortion Matrixof Group Of Pictures( GoP)(as in an MPEG-2 stream) It is calculated under thethat once a specific P frame or I frame is lost, all the depending frames in theGOPre replaced with the latest successfully dd frame This assumption makes the (DMmodel unsuitable, because today all video decoders tend to mitigate the error propagationreducing the distortion after a single loss

The GOP structure of video k is described by theGOP length Lk and the number of B-frames Bk between two I or P frames, For example, withBk=2 the goP structure wiBPBB The obtained distortion matrix as inis shown in Figure 8R:D, DR DB, DR DBDBDBDB DR,1: DADADA DE, DB, DA DA, DH/// DA DR DE D' DPDB DBFie 8 DM for Lk =9 and bk=matrix are the mse values observed

Itimedia TrafficWireless and Satellite Networksreplacement frame for every row of the matrix R is a frame from the previous GoP that islost From this matrix, the resulting distortion for any possible losspattern can be determined The total distortion for the Golindividual frame loss distortions This matrix can be determined during the encoding of thedeo the number of columns of the distortion matrixponds to the goP length Lbased"methods, because they concern the distortion introduced at the frame level; othedistortiell-known in the literature and theyas"pixel-based" methods In particular, the block-based approach generates and recursivelypdates a block-level distortionch frame 32, 33, 34

Nevertheless, since inter-inherit errors propagated from prior frames In contrast, pixel-based policies estimate thObviously, on the other hand, the complexity goes increasing An innovative approach haspixel-basis is calculated by exhaustiveimulation of the decoding procedure and averaging over many packet loss patternsconsideredhe ropebe preserved, without losing the optimality of the distortielv calculates the first and secondts of the decoder reconstruction ofh pixel, whurately taking into accorcluding erropropagation and concealment4 MPEG Standard""GOP Structure-lenght 15 framesFig 9, GOP StructureMPEG is the acronym of Moving Picture Experts Group, a working group which has thand audio encoding standards [38]MPEG video traffic is characterized bystant transmission rate of twops of pictured and 15 frames per GOP Since the number of bytesframe is dependent upon the content of the video, the actual bit rate is variable over timeHowever, the MPEG video supports also the constant bit rate( CBr)mode There are threetypes of framesI-Frames(intraframes)-encoded independently of all other frameP-Frames(predictive frames)-encoded based on immediately previous I or PframesB-Frames(bidirectionally predictive)-encoded based on previous and subsequent

Many works infaced the problem of analyzing and describing the structure oftrying to find a way to emulate them through statistic andtochastic streamators, preserving the properre of the originalstream In [39] the inputDependent(LRD) and short range dependence (SRD) models, Simulation results werefound to be better than those of a self-similar process when the switch buffer is relative'Group of Pictures'(GOP) level presented in [40] is a Markov chain model based on thef eliminating the cyclical variation in the MPEG video pattern, but at thethe time scaleGOP has duration of a halfdered long for high speed networks Of particular intereststhe framesize distribution and the traffic correlation the frame sizhas been studied in many existing works, Krunz [41] proposedMPEG virelated component is introduced in the modeling of I frames,effectsizedby a modulated process in which the localare modulated by an Auto-Regressive(AR) process thatlated random process withmodulatedwith log-normal marginal This modelseveral randomomplicatingthe adaptive source is modeled by means of a discrete-time queuing system representingvirtual buffer loaded by the video source when itsbased on the Switched Batch Bernoulli Process (SBBP) that has been demonstrated to betable to model an MPEG video source; in fact, being a Markov modulated model, anBBP is able to capture not only the first-order statistics but also the second-order onencept of GOP-radeling, based on the discretisationethod originally proposed in [44] for the wireless channel study After the GOP-rate trendas been analyzed for the whole duration of the stream, it has been discreMPEG algorithms compress dform small bits that can be easily transmitted and thento another, instead of each entire frame

The video information is then encoa technique called Discrete Cosine Transform (DCT) MPEGa typeperceptible to the human eye The major MPEG standards include the foMPEG-1: The mmmon implementations of the MPEG-1 standard provide a videoresolution of 352-by-240 at 30 frames per second (ps) This produces vidslightly below the quality of conventional VCR videosMPEG-2: Offers resolutions of 720x480 and 1280x720 at 60 fps, with full CD-qualitydio This is sufficient for all the major Tv standards, incluuseda 2 hour video intoWhile decompressing an MPEG-2 data stream requires only modestodingMPEG-2 formataires significantly

Itimedia TrafficWireless and Satellite NetworksMPEG-3: Was designed for HDTV but was abandoned in place of using MPEGHDTVMPEG-1 and MPEg-2 andQuickTime technology Wavelet-based MPEG-4 filese smaller than JPEG or QuickTime files, so they are designed to transmitrrower bandwidth and can mix vidh text, graphics an3-D animation layers MPEG-4 was standardized in October 1998 in theocument 14496dpletely describing multimediageted to a specific applicaMPEG-21: Includes a Rights Expression Language(REL) and a Rights Data Dictinlike other MPEG standards that describe compression coding methods, MPEG-21describes a standard that defines the description of content and also processes forccessing, searching, storing and protecting the copyrights of contentJPEG 2000DigitairelessvideoconferencingInternet(RTP, Streaming, etcFig 10

Multimedia Broadcast Satellite Scenariointeractivity between users and network(see Figure 10) Thanks to this interactivitye able to manipulate what they receive on their terminal requiring to the network whamake retry to a symmetry network Then it way to exploit the network that pass from aney want The interactivity has changed thetellite cofns that, ofthe most oriented broadcast medium Originally,PSTN ISDNNowadays the mosttechnology used for the satellite interactivitllite one, in fact, the new standards fo

Digital Videotellite communications propose a return channel via(RCS)like the DVB-RCSfor this chotechnology for simplicity of management It is very sire a single box in which alncentrates without have thebjects Another importantn regards the increased traffic in terrestrial networks thatoviding suchblockste available for the applications reaching a 2 Mbps against the few hundredkbps of the terrestrial solutions An example of thisailable bansfer application that for a 100 Mbyte file it will need about 7 minutes against thebout 3 12 hours of the terrestrial lines that operates a 64 kbps

There is also an advantageboth for the users and the operators, thatThisprodand in network management by operator, infact, the terrestrial infrastructure is often not controlled by the same operator as for satellite,rtainly true whenUntil 60 years ago, each individual communicated with about 100 other persons, of whicht livedhisindividual communicates with about 500 other individuals, of which 80-90 percent do negencmmunications systems and by the integration of different technologies of communicationhat have allowed to reachtechnolohas allowed the integration of different devices that originally worked separately asmputers, TVs, telephones Thanks to the Inttechnolpid technological advances will allow in a next future to have a exchange of informationFig 11 DVB Satellite Terminalproblem of last mile providing connections to thosebe possible because a large investment is required to bridge sethe localange and the customers Satellite, thanks to its broadband nature, is able to provide

Itimedia TrafficWireless and Satellite Networksconnection in the rural areas and isolatedinvestment of otherinal(see Figure 11)ey are able to receive satellite information Moreover, thanks to the new RCS standard thek interactivity exploiting the return satellite channel thalis faster than terrestrial one which is based on telephony network thatmited inoviding high bit rates to subscribers It will be necessary to perform a fiber cabling alsobetween local exchand subscribers in order to make faster the connectianies to perform a fiber cabling in orderguarantee high bandwidth to each subscrilIt is clear that the increase of multimediaovide highbandwidth to the users by operators and in this context the use of satellite platfpresents an optimal solution in terms of costs Moreover the satellite segment guaranteesgration with all communication technologies51 Digital Video BroIn the mid-eighties, the ability to transmit digital images was stillte and it was thoughtthat it was neither technically nor economicallyas the high bit rate required, especially for the transmission of digitmotion(from 108 to 270 Mbps) The most important issue was to improve the quality of thenineties has completely changed: the creation of efficient compression algorithmsresultedesult, the birth of the standard Pegfor fixed and then theMPEG standard for moving images; thanks to MPEG compression the amount of dataequired for transmission of digital images up to debetween 1

5 and 30Tbps has been reduced drastically, depending on the resolution chosen and the type ofnents of suppliers of services, but only if they had adopted a common standardto the DVB Project [45] DVB, short for Digital Video Broadcasting, isused with reference to digital television services in accordance with the standarddped by a consortium of 300bthan 50countries, operating in different sectors: from production to broadcast TV of television setsby the rules of the frequency spectrum to the study of protocols for access to a network Thebers of thisrk

with the DvB pelop a set of standards,technical recommerailable to the various manufacturers onceished by ETSI(EuropeanTelecunication Standards Institute), and then madble to all So thanks to thesestandardlufacturers can create interoperable DVB systems and can also easilydapt to different transmission channels (satellite, terrestrial, cable, etc ) Althoughborn for the European landscape, the Dvb platform is beginning to be accepted asd television programs based on DVB standardrrently operating in North and South America, Africa, Asia and Australias is based on the standard Iso 13818 encoded MPEG-2 and multiplexingMPEG-2 satellite, cable anterrestrial repeaters, and as transmitting the information system, the program guide, efDVB allows the broadcasting of"data containers"that can include digital data of any type,

Itimedia TrafficWireless and Satellite NetworksOver the years have developed many models, more efficient and less expensive, able toodels of measurement is reported in the Figure 1 below They are divided substantiallyCalibrationeg listening only and conversationaladditionalSignal basedasurementsComparisonestimationrating modelsMeasue ccmbined effect ofPredict combined efect ofdividually measured impairmentsFig

1 Overview of models of measuren21 Subjective quality assessmenttechnique for obtaining quality ratiItallv, thementsa room where there are a certain number of persons(about 15-30)that they havea set of video streams and, then, they have to give a rate based on own qualityperception On the basis of all rates given by all subjects included in the experiment it isabjective evaluation determined by the subjectivityability of the involvehis test is affected by the personal opinion that cannot be eliminated In ordtype of problem the experiect a set ofalso the environment usedthat is a number that results fr

4Digital Videoariety of subjective testing methods Int methods exist for givconditieat are used for qualityrecommendations [10, 11, 12, 13, 14They suggest standardteria for the selectionobservers and testsment procedures, and data analysis methods, Rded testingexist as Absolute Category Rating(ACR), Degradation Category Rating (DCR)Corder to obtain a more realistic statistical score After each video presentation,ked to judge its overall quality using the rating scale shefigExcellentGoodFairBadFig 2 Subjesessmenttime-limited After choosing their qualityd to confirm their choice using an 'OK button Furthermore,liminated thepossibility of missing ratings in the test

After each video sequence and after to have gavuence is presented The tesadhered to the latest findings and recommendations for best practice from the Video QualityExperts Group(VQEG), a technical body supporting ITU standardization activitiesuse of this type of test has some disadvantages, first of all, the result of the test deple attributes like experience, the mood, the attitude and culture, then,and impractical if you want to do frequently because of the number ofbjects and tests are neto give reliable results anbiective assessmentinvaluable toolsaluating multimedia quality Their main shortcrequirement for a large number of viewers, which limits the amount of video material thatbe ratedt of time Nonetheless, subjebenchmark for anmethods aasible during the design of a network These are limited,Ihese problmethods that allow the calculation of values that represent the different combinations of

Itimedia TrafficWireless and Satellite Networksfactors of damage that could affect the networkpurpose of these methods is tovalues The ITu is proposing a method of objectilatic and repeatable, whichccount the perceived quality Diffethe analysis of decoded video, weistinguish data metrice the fidelity1gontain, and picture metrics, which treat the video datainformation and the encoded bim directly without fully decoding the videoFurthermore, metrics can be classified into full-reference no reference and reducedeference metrics based on the amount of reference information they requireThe most known video quality metric is called Peak Signal to Noise Ratio(PSNR) thatthe original video The other popular metric is the classical Mean Square Error(MSE) ist by which an estimator differs from the truethe quantity being estimated3 Multimedia over wireless networksNowadays multimedmunication over wireless and wiredgrowing up

In the past, many applications were used for video downloads, while now thetake up the share of all traffic on the Internet Most mobile devices can actively downloadnd upload photos and videos, sometimes in real time In additionand the enhanced televinvironment: the electronics manufacturers, the computer industry and its partners aredistributing audio and videoocal-WiFi networks to monitors and speakers around theNow that the analog-to-diplete, the"all media overution is taking place, with radio, television, telephony and stored media all beingIP wired and wireless networks Figure 3 shows an example of differentThe growing and the emergInfrastructures, like the internetwireless networks, enabled theiferation of the above mentioned multimediapplications(music download to a portable device, watching TV through the Internetlaptop, viewing movie trailers posted on the web through a wireless link) New applicatiending VolP to an apparentlpparently conventional set top box or sending music overpparentlystereo amplifier The exposed applications include a bigvariety of new multimedia related services but, unfortunately, the Internet and the wirelessnetworks do not provide full support for multimedia applications The Internet and wirelesse stochasuenced by many factors, howevariations in network conditions caapplicatandlead tobecause multimedally delay sensitive, bandwidth intense and loss tolerant The thery, communicationprocessing may not be directly applied to highly time-varying channel conditions and, as

uence, in recent yoe area of multimedia communication and networking hasvery active and challenging research topic, but also asthe definition of new fundamental concepts and algorithms that differ from thoseconventional signal processing and communication theorySAAudio systemFig 3 Different Wireless IP-based multimedia applications (STP-Set Top Box for onmand video, HDTV-High Definition TV, DMAfor multimediane extensions, PVR- Personal video recorder etIt is clear that Best-Effort(BE)IP networks are unreliable and unpredictable, expewireless networks, where there can be many factors that affect the Quality of Service(Qothate of a transmission system via parameters that reflect itstransmission quality, such as delay, loss and jitter In addition, congested networkvideo Further, there are strict delay constraints imposed by streamed multimedia traffic If apacket does not arrive before its"playout time", the packet is effectively lost

Packetlue to inter-frame dependencies A slightly degraded quality but uncorruptedstream is less irritating to the user than a corrupted stream Controlled video qualitydaptation is needed to reduce the negative effects of congestiothe stream whilstproviding the highest possible level of service and quality Thedependently in the network, so that the receiver is able to decode and play baclof the bit stream that are alreadyThe transmitter continues to sendsets while theItaneously plays back other alreadParts of theThe philosophy of playing back received packets allows the reduction of the delay betweenthe transmission time instant and the moment at which the user views thecontent Having a low delaysuch kinehereconferencvideodemand architecturelThe transmission of multimedia content can be categorized into three main classes: unicast,multicast and broadcast, depending on the number of senders and rece

Itimedia TrafficWireless and Satellite Networkstransmission connects one sendernnection,unicast is that a feedback channelbe established between the receiver and thetransmitter so theconditions which can be used accordingly by the transmitter to changparameters In a multicast communication the sendereivers thattilizationmiddle nodes(it is obvious that in a multicast communication the sender cannot openn toward a specificIn a broadcast transmission the sender is connected toall receivers that it can reach through the network(anbe different for every receivetidelay is to use feedback mechanisms to adapt the output bit rate of the encoders, which, inturn addeo quality, basedstate of the network Several bit rate control mechanisms based on feedback have beenresented in the last few years As the Real-Time Control Protocol(RTCP) providestransmission of video over IP networks However, the network-level Qos parametersovided by rICht video content based and it is difficult tothe quality of theeceived video stream from this feedbackthe following paragraphs, multimedia transmission techniques will be deeplyof the perceived content31 Backtransmission of video over wireless media is becoming very popular for theariety of applications and networks

The advantages of thesevident, but the transmission rate will alwatroduced by physical layer So, having the possibility of compressing the video beforconstraintsthese reasons, video decoder must tolerate dpacket losses: the standards in video coding (like MPEG-4 and H 264/AVC)[16, 17, 18, 19]aretodaytheir capability to adapt to these environmentsThese standards, like thekIuse a hybrid coding approach TheMotion cotion(MCP) is combined with transform coding of the residualnts inr to obtain a codec thatprevious paragraph, video communications can be categorized into unicast, multicast andbroadcast servih diffpeculialike thePrFigure 4 illustrates a generic scheme of a video transmission system, including the mainents that go from the trater device to the receiver device Data compressiignificant redundancy (in time and in spaceed packets significantling to Figure 4, thgeneratontaining the compressed video stream, which is storedin the encoder buffer before the transmission In general, the transmission system damages

quantitatively or qualitatively)individualits, also introducing some delay; then, thethe encoder and by the wireless channela coded video streamconsiderAbstraction Laver Units (NALU in H 264 The AUS/NALUs can be labeled as data unitpecific information(as in MPEG, where they are labeled on the basis of relative importancefor video reconstruction) On the other hand due to spatial and temporal prediction thependent compression of data units cannot be guaranteed without significantly losingmpression efficiencyCodng ControlVideo MultiplexTransmissionVideo coderSourceadeo MultiplexDecoderHiding Decease hiFig

4 Blocks diagram of a generic audio/video transmission systemErrors introductioits effects are considerably different in wired or in wirelessnetworks, because of the different phenomena that impact on the medium For wirelessnetworks, fading and interference cause burst errocongestion cawireless networksstems are able to detect thein a packet on physicaler and the losseorted to higher layers These techniques usually use CyclicRedundancy Check( CRO) mechanisms, By consequence the video decoder will not receivehe entire bit-stream Intermediate protocol layers such as User Datagram Protocol (UDP)ts In fact, video data packeonsidered lost if their delayfixed andFigure 5 illustrates a typical simplified version of an end to end video system when the MCPpresents the time, S, is a single vk packet Pa, C(withThe predi nd me packet is forled to the normal decoder operation such as Entropynformatiostream to reconstruct the current frame S -1 After that, theis forwarded to the displabuffer and also to the reference frame buffer to be used inP process to reconstruct

Itimedia TrafficWireless and Satellite Networks9s(Cme)5 End to end videothe foter-coded frames (ie the frame st frame) When the packet Pr is lost, ie atne reference time t, Ct=0, the Error Concealment (EC) is necessary to be enabled, so thedecoder just avoids the decoding operation and the display buffert updated, so thed frame is still sl In this case thewill understand that there has been a lossce the continuous display update is not maintained

Also the reference frameresult of this data lIn case of successful reception of packet P+l, the inter-coded frame SHI, reconstructed at thedecoder, will in general not be identical to the reconstructed frame Sri at the encoder side,is reason it is obvious that the loss of a single packet Pr affects the qualityall the inter-coded frames: St+l, 5:+2, Sp+3, etc This phenomenpredictive codingscheme and is called error propagation If predictiveve coding is appliedin the spatial andpatio-temporal errCP technique makes the reconstructed frame S, not only depending on the actualhannel behavior C but also on theAfter these considerations, remarking that the error propagation has direct consequences onperceived video content, it is preferable that a video coding system pited to the reduction of the visual effectsreceived framesalgorithm or device for the minimization of the error propagation effectGenerally, in MCP coding, the video content that belongs to a single frame is not encodedindependent The features that are employed in these systems to correct thError Correction(FEC)and Backward Error CorrectionEC)or any combinations of those It is very important, at the receiver size, that thed missing video content is observed, localized and, if possible, eliminatedodern video coding systems also have the capabilitythe video encoderabout the loss of the vintent, so the encoder can adapt itself to enhance the

ty, The macro-blocks assignments, error control methods and feedbackson, for example, must be used if we desire a robust, efficient andAknown in literature, if Shannon, s separation principle [20] is observed, the main goal inoid losses in video transmissionsseveralbe impossible and the features like channel loss correction, detection andble errors their minimization reduction of distorted frames impactbecome essentials When we are dealing with 'no-wired systems and Qos guarantees mustprimarythat should be provideded by the video coding standards, such as H 263[ 21, MPEG-4nd H264, beyond the syntax and the semantics of the transmitted data, specify the decoderbehavior in case of reception of an error-free bitstream, so the deployment of video codingstandards still probitstreamthis way, a processing algorithm may be better than anotherlexity or quality of thecontent

In the lastandards, like H 26MPEG, MPEG-2 23chile the latter ones(starting from the H263 [251), heavily influenced the applications, with a lot oftime,the new emerging standardMPEG-4 Advanced Simple Profile(ASP) introduced a radically differenersible Variable Length Coding (RVLC) and Rynchronization Markers(RM)were introduced [26] Figure 6 shows the evolution in time ofn standardsVideo Compression StandardsH261H26H26MPEG 1MPEG 4MPEG 2standardsAs exposed in previous paragrdeal channelinsert additional and redundant data to the information sequehannel errorsbe detected or corrected Shannons channel coding theoremhat if there is