Spectrum Management. Digital Audio Broadcasting. Content Protection. Video Streaming. Quality of Service

Transcription

1 Wecome 3 Spectrum Management Impementation of the Digita Dividend technica restraints to be taken into account Jan Doeven, KNP 4 Digita Audio Broadcasting The evoution of DAB Frank Herrmann, Larissa Anna Erismann and Markus Prosch, WordDMB 12 Content Protection HDMI & HDCP the manufacturers perspective Dietrich Westerkamp, Thomson, EICTA HDTV Issue Manager HDCP the FTA broadcasters perspective Jean-Pierre Evain, EBU Video Streaming Mutipe Description Coding a new technoogy for video streaming over the Internet Andrea Vitai, STMicroeectronics 35 Quaity of Service Network structures the internet, IPTV and QoS Jeff Godberg and Thomas Kernen, Cisco Systems EBU TECNICAL REVIEW 1

2

3 2007 The best of Wecome to the Best of 2007 seection of artices which have previousy been pubished eectronicay in EBU Technica Review at htm Since 1998, the Technica Review has been pubished on-ine, four times per year. This venture has been very successfu because it has expanded the readership. It has been very gratifying to meet peope who have discovered the on-ine version. Some of these new readers are not directy connected with the EBU but they work in reated areas, such as suppiers of hardware or software to the broadcasting industry. Our eectronic pubishing service incudes an archive of the Technica Review, dating back to You can access these archive artices by cicking on the Archive / Thematic Index ink, in the navigator frame to the eft of the screen. This archive contains a weath of exceent artices on many different topics. Recenty we added a Hot Topics section to the on-ine version which brings together reevant artices from the archive on topics such as HDTV in Europe and Broadcasting to Handheds. The on-ine version aso incudes a ist of abbreviations used in the Technica Review over the past 15 years or so. New terms are continuay being added to the ist which you can downoad as a PDF fie by cicking on the Abbreviations ink in the navigator frame. in terms of portabiity, weight, readabiity, quaity, price and (ast, but not east) battery ife! As there is sti considerabe vaue in paper pubication, the EBU decided to re-pubish some of the artices in an annua printed edition of EBU Technica Review. Whereas the onine version is avaiabe ony in Engish, the annua edition is aso avaiabe in French. If you enjoy reading the artices in this pubication, remember to consut the on-ine version of EBU Technica Review at Finay, do not forget to give this URL to your friends and coeagues! Lieven Vermaee Director EBU Technica Department At the beginning of the year 2000, the EBU abandoned the printed version of EBU Technica Review. Nevertheess, it was recognised that eectronic pubishing coud not entirey repace hard copies. This state of Nirvana wi not arrive unti we have computers that can match this paper pubication 2007 EBU TECNICAL REVIEW 3

4 SPECTRUM MANAGEMENT Impementation of the Digita Dividend Jan Doeven KPN technica constraints to be taken into account At the RRC-06, a new Agreement and associated frequency pans for digita broadcasting and anaogue TV broadcasting during the transition period were agreed (GE06). The next step is impementation of the new Agreement. Broadcasting organizations, network operators, spectrum user forums and others have announced their opinions on the use of Bands III, IV and V. A term often used in reation to the impementation of the new Agreement is digita dividend. This artice describes the technica constraints to be taken into account when using reeased spectrum for severa digita dividend appications. 1. Introduction At the RRC-06, a new Agreement and associated frequency pans for digita broadcasting and anaogue TV broadcasting during the transition period were agreed (GE06) [1]. The next step is the impementation of the new Agreement. Broadcasting organizations, network operators, spectrum user forums and others have announced their opinions on the use of Bands III, IV and V. A term often used in reation to the impementation of the new Agreement is digita dividend. There may be many meanings of the term. For the countries in the European Union, the definition used by the Radio Spectrum Poicy Group (RSPG) and the European Commission is most reevant. Digita Dividend is, according to the RSPG, to be understood as the spectrum made avaiabe over and above that required to accommodate the existing anaogue teevision services in a digita form in VHF (Band III: MHz) and UHF (Bands IV and V: MHz) [2]. It shoud be noted however that existing anaogue teevision aso makes use of Band I (47-68 MHz) and, after digita switchover, Band I spectrum coud be considered as digita dividend too. Furthermore, Band III is aso panned for T-DAB and many existing T-DAB services aready make use of Band III. In addition, in a number of countries, non-broadcasting services make use of Bands III, IV and V. Many possibe appications of the digita dividend are under discussion. In its Communication on EU spectrum poicy priorities for the digita switchover in the context of the upcoming ITU Regiona Radiocommunication Conference 2006 (RRC-06) [3], the European Commission identified three categories: 1) Spectrum needed for the improvement of terrestria broadcasting services: e.g. services with higher technica quaity (notaby HDTV), increased number of programmes and/or enhancement of TV experience (e.g. muti-camera anges for sports, individua news streams and other quasi-interactive options); 2) Radio resources needed for converged broadcasting services which are expected to be primariy hybrids of traditiona broadcast and mobie communication services; 3) Frequencies to be aocated to new uses which do not beong to the broadcasting famiy of appications. Some of these potentia new uses 4 EBU TECNICAL REVIEW 2007

5 SPECTRUM MANAGEMENT of the spectrum dividend are future services and appications which are not yet marketed and others are existing ones which do not operate yet in these frequencies (e.g. extensions of 3G services, short-range radio appications). This artice describes the technica constraints to be taken into account when using reeased spectrum for severa digita dividend appications. Considerations on the use of the digita dividend are aso described in [4]. 2. Size of the digita dividend 2.1. Layers A term often used when considering nationa input requirements and resuts of RRC-06 is the number of ayers. A ayer is not defined in the GE06 Agreement, nor was it defined at RRC-06, but for most European countries it may be described as a set of channes which can be used to provide fu or partia nationwide coverage. The number of ayers depends, among others, on the geographica situation, the eve of accepted interference, transmission and reception characteristics and the way an Administration composes its ayers out of the avaiabe Pan entries. Administrations submitted their T-DAB and DVB-T requirements before RRC-06. Fufiing these initia requirements woud, in some areas, have required ten times the band capacity and, in most areas, two or three times. In defining input requirements, Administrations took into account their ong-term broadcasting needs, their rights concerning use of other primary services operating in Bands III, IV or V (if Guidance for number of ayers Band III any) and maybe, in some countries aso, possibe future use of other appications. However, as the panning process at RRC-06 aowed Administrations to make input requirements ony for T-DAB or DVB-T, other possibe appications needed to be described as broadcast requirements. Another eement in defining input requirements was the wish for a Administrations to have an equitabe access to the frequency bands. Therefore the T-DAB and DVB-T input requirements do not aways necessariy represent the current minimum market requirements. Furthermore, it shoud be noted that a nationa requirement may seem unreaistic from a frequency-panning point of view, or even from the point of view of a neighbouring Administration, but coud be poitica reaity in a country. During RRC-06 there was a strong pressure on Administrations to reduce their requirements in accordance with the foowing guidance (see the tabe beow). Most European countries were successfu in achieving the above-mentioned number of ayers. In most countries there are four anaogue TV services and these can in genera be accommodated into one DVB-T mutipex for which one DVB-T ayer is needed. However countries with five or more anaogue TV services and using DVB-T with a robust moduation, may need two DVB-T mutipexes and thus two ayers for broadcasting their existing anaogue TV services in digita format. For a successfu introduction of DVB-T, more mutipexes are needed than the number of channes containing the current anaogue TV programmes (see Section 6.1) but, foowing the RSPG definition, in genera out of the eight to nine achieved DVB-T ayers, six to eight DVB-T ayers Band IV/V and the three T-DAB ayers coud be seen as digita dividend (Fig. 1). Anaogue TV transferred to digita Figure 1 Band III, IV & V spectrum Digita dividend according to EU definition 2.2. Frequency bands Band I (47-68 MHz) was not panned for digita broadcasting at RRC-06 and is reguated by the revised Stockhom Agreement [5]. The band is not incuded in the RSPG definition of digita dividend. However, after anaogue TV has been switched off, it may aso be considered for new appications, taking into account that there are aready non-broadcasting services in a number of countries. Band I is ess attractive than Bands III, IV or V for many services due to: its ong waveength, and therefore arge antenna dimensions; its susceptibiity to ionospheric interference from the Sporadic E-ayer; the high eves of man-made noise at these frequencies [6]. In genera, not much interest has been expressed for Band I. Currenty some DRM (Digita Radio Mondiae) experiments take pace in this band. Band III ( MHz) has been panned for T-DAB and DVB-T. A number of countries are considering impementing DVB-T ony in Band IV/V, and to use Band III excusivey for T-DAB or mutimedia appications making use of a T-DAB based system. There is currenty no interest in appying new non-broadcasting services in this band. T-DAB DVB-T DVB-T For digita dividend appications, the band is considered as 1 st and 2 nd category (see Section 1) EBU TECNICAL REVIEW 5

6 SPECTRUM MANAGEMENT Band IV/V ( MHz) is subject to most of the discussions on digita dividend, covering a three categories (see Section 1). In addition to broadcasting, the UMTS obby sees it as an attractive band for mobie communication systems. 3. GE06 Agreement The Pan entries of GE06 wi ony become fuy avaiabe after anaogue switch-off. The European Union proposes to switch off anaogue TV before 2012[7]. According to the GE06 Agreement, anaogue TV wi have no right of protection after 17 June 2015 (and, in some African and Midde East countries, after 17 June 2020 in the case of VHF transmissions). The GE06 Agreement offers two options to achieve fexibiity in the appication of Pan entries (Artice 5): Different characteristics of a Pan entry can be appied as ong as the conformity check is fufied. The main criterion is that interference from the appication is not more than that of the Pan entry. This mechanism can for instance be appied to convert a Pan entry into a Singe Frequency Network or a different reception mode (see Section 4); Aternative appications of a Pan entry (that is other than DVB-T or T-DAB) are possibe in the Broadcasting, Mobie and Fixed services if three conditions are fufied: band aocation in the Radio Reguations to the reevant service; not exceeding the spectra power density of the associated Pan entry; not caiming more protection than afforded to the associated Pan entry. A more detaied description of the options for achieving fexibiity is given in [8]. In addition, the GE06 Agreement contains a procedure for modification of the Pan (Artice 4). Under this procedure, the agreement of a potentiay-affected countries is needed to make a change to a Pan entry. The Artice 4 procedure aso needs to be foowed in cases where services other than broadcasting, which have co-primary status, are introduced or modified. Depending on the impact on the GE06 Agreement, two uses of the digita dividend can be distinguished: Appications making use of Pan entries which require no or imited modifications to the GE06 Pan; Appications making use of a dedicated sub-band with the consequence of considerabe modifications to the GE06 Pan. 4. Appications making use of GE06 Pan entries The GE06 Agreement has harmonised panning parameters for use of the MHz band by T-DAB and DVB- T and the MHz band by DVB-T. T-DAB has been panned for mobie and portabe reception, DVB-T for rooftop and portabe reception Reception mode Each Pan entry has a specified reception mode. The most used for DVB-T are a set of characteristics for rooftop reception or portabe outdoor reception. The atter term stands aso for portabe indoor or mobie reception at a ower coverage quaity. Fig. 2 shows the specified DVB-T reception mode for the European countries. A transmission based on a Pan entry specified for rooftop reception can be used for portabe reception if a reduced coverage area is acceptabe. If it is not, a dense Singe Frequency Network (SFN), that fufis the conditions of the conformity check of Artice 5 of the GE06 Agreement, is a possibiity to improve coverage. It may aso be necessary to seek internationa agreement for modifying the Pan entry with a higher power by appying the Artice 4 procedure of the GE06 Agreement. If a transmission based on a Pan entry that was specified for portabe reception is used for rooftop reception, a arger coverage area wi be obtained and there may be an overap of (rooftop) coverage of two or more adjacent transmitters. In practice, portabe coverage may be restricted to buit-up areas. In the surrounding rura areas, the wanted fied strength is ikey to be sufficient for rooftop reception but the received interference eves associated with the portabe Pan entry may be too high for fu rooftop coverage of the area. There is ikey to be more than one possibe wanted transmitter, because of the overap of coverage areas. In some cases, instead of directing the rooftop antenna towards the transmitter giving the highest signa strength, a better signa-to-interference ratio may be obtained by aigning the antenna on another transmitter. In some areas, a very directiona (and hence, a much more expensive) rooftop antenna may be needed. It may aso be necessary to optimize the transmitter characteristics, or the SFN, taking care that the conformity check of Artice 5 of the GE06 Agreement is fufied. The most-used basic characteristics of the DVB-T reception modes in Band IV/V are: Reception mode Rooftop Portabe Capacity 24 Mbit/s 16 Mbit/s Required fied strength 56dBµV/m 78dBµV/m 6 EBU TECNICAL REVIEW 2007

7 SPECTRUM MANAGEMENT 5.1. Repanning Band IV/V A new non-broadcasting appication needs to be agreed by a potentiayaffected countries in accordance with the Artice 4 procedure of GE06 and has to be incorporated in the List of Annex 5 of GE06 in order to be protected from GE06 Pan entries and further modifications of the GE06 Pans. In order to obtain agreement, the new appication may be subject to restrictions because Pan entries of other countries need to be protected and interference from Pan entries of other countries accepted. Figure 2 Reception mode as specified in GE06 The fied-strength requirements for handhed reception in Band IV/V range from 85 to 107 dbµv/m, depending on the moduation and reception conditions [9] and are higher than for portabe reception. A transmission based on a Pan entry for portabe reception can be used for handhed reception under comparabe conditions to those indicated above for the case of a Pan entry for fixed reception being used for portabe reception Different network topoogies If, in a given area, the network topoogy of one or more of the mutipexes is different to that of the other mutipexes (e.g. if dense networks are used for some mutipexes), adjacent channe interference may occur around non co-sited stations. Such interference may occur on the first, second and even third [provided by the EBU] adjacent channe on both sides of the wanted channe. Adjacent channe interference is a oca probem. A possibe soution is coocating fi-in transmitters at the site of the interfering transmitter. The question wi arise however who wi have to pay for these provisions. 5. Appications making use of a dedicated subband For appications with up-inks and different channeing schemes, dedicated sub-bands are considered [2][18] e.g. for UMTS. In the case of UMTS, an aocation to Mobie services in the Radio Reguations needs to be agreed at ITU WRC-07 or WRC-11 and a sub-band woud be required from which the GE06 Pan entries are deeted. In genera in the GE06 Pan, the frequencies at a given site or in a certain area are scattered over the whoe band. A sub-band for Mobie services, such as UMTS, coud therefore affect a DVB-T ayers (for the meaning of ayers see Section 2.1) as it woud create hoes in the ayers (areas not covered because the frequencies are no onger avaiabe due to the sub-band). The remaining part of the band wi need to be re-panned in order to obtain the origina envisaged DVB-T coverage, with a reduced number of ayers, by appying the Artice 4 procedure of the GE06 Agreement. This re-panning means in practice a re-doing of the GE06 Pan. However Pan entries of other countries need to be protected and interference Launched Preparing Figure 3 European DVB T map (from DigiTag) 2007 EBU TECNICAL REVIEW 7

8 SPECTRUM MANAGEMENT from those Pan entries of other countries accepted. The re-panning process is ikey to be compex and time-consuming and it is not guaranteed that the origina coverage can be repaired. DVB-T has aready been introduced in 14 European countries (see Fig. 3) and, by the time the process is competed, many more DVB-T transmitters wi be in operation. A transition from the origina GE06 Pan to a re-panned GE06 Pan wi be necessary. 5.2.Guard bands In order to avoid interference between upink transmissions and adjacent (downink) broadcast transmissions, Abbreviations guard bands are needed. The width of a guard band depends on many factors and, according to ongoing studies in ITU-R and esewhere, may be more than 10 MHz. Aso a guard band is needed between the upink and the downink sub-band. The tota guard bands and thus the unused spectrum may add up to severa DVB-T channes. 6. Spectrum use Bands III, IV and V are the ony avaiabe bands for obtaining wide-area DVB-T and DVB-H coverage. Foowing an Opinion of the Radio Spectrum Poicy Group of the European Union on the introduction of mutimedia services[10], CEPT has been mandated to identify appropriate technica CEPT Conférence Européenne des Postes et Téécommunications (European Conference of Posta and Teecommunications Administrations) CRT Cathode Ray Tube DAB Digita Audio Broadcasting (Eureka-147) DRM Digita Radio Mondiae DVB Digita Video Broadcasting DVB-H DVB Handhed DVB-T DVB Terrestria EU European Union GE06 Geneva Frequency Pan of 2006 ITU Internationa Teecommunication Union ITU-R ITU Radiocommunication Sector RRC (ITU) Regiona Radiocommunication Conference RSPG Radio Spectrum Poicy Group SAB Services Anciary to Broadcasting SAP Services Anciary to Programme-making SFN Singe-Frequency Network ST61 Stockhom Frequency Pan of 1961 T-DAB Terrestria DAB UHF Utra High Frequency UMTS Universa Mobie Teecommunication System VHF Very High Frequency WRC (ITU) Word Radiocommunication Conference and reguatory parameters for opening up the band 1452 to MHz to aow fexibe use by a wide range of mobie mutimedia technoogies. However the propagation characteristics and the width of this band (25.5 MHz) are in genera not adequate to pan nationwide coverage in each of the European countries, even if a 5 MHz DVB-T or DVB-H bandwidth is chosen. UMTS services can be operated in severa bands and a series of possibe extension bands have been identified incuding Band IV and V[11] Broadcasting use of Band IV and V In order to motivate consumers to buy a digita receiver for terrestria services, an attractive broadcast package needs to contain 20 to 30 popuar programmes. Such a number is aso needed to provide better competition to sateite and cabe deivery. A arge number of programmes that are of high individua interest to ony a few peope can best be deivered by means of on-demand services: for exampe, via UMTS in the case of reception on sma screens. This coud be faciitated by means of a common Eectronic Service Guide (ESG). These on-demand services coud incude pubic and commercia programmes received outside the nationa territory, for instance by traveers and tourists wishing to receive their homeand programmes. The number of ayers which can be provided by GE06 is very arge and significanty increases the spectrum usage as compared to ST61; in fact, this usage exceeds the theoretica capacity of the frequency bands, at east for the technica conditions used at RRC-06. These extra ayers have been achieved at the expense of accepting higher interference eves which may resut in ower quaity services and/or reduced coverage areas. To overcome these difficuties when impementing the Pan and in order to provide reiabe 8 EBU TECNICAL REVIEW 2007

9 SPECTRUM MANAGEMENT needs to be transmitted in parae to DVB-T mutipexes. services, it may be necessary to depoy additiona transmitters and additiona frequencies. In order to provide an acceptabe video and audio quaity on conventiona dispays, three to four programmes can be accommodated in a mutipex for portabe reception (16 Mbit/s) and five to six programmes in a mutipex for rooftop reception (24 Mbit/s). The average data capacity aocated to each programme coud be from 3 to 4 Mbit/s depending on the DVB-T variant used and depending on the statistica mutipexing, if used [12]. It shoud be noted that quaity requirements need to be increased with the advant of fat-pane screens. These kinds of screens are very popuar and are, or wi be, used soon in many househods. EBU investigations have shown that fat screens are more sensitive to artefacts and, for a good picture, require about twice the bitrate needed for Cathode Ray Tubes (CRTs) [13][14]. The video compression system MPEG-4 wi enabe a ower bitrate compared to MPEG-2, whie maintaining the same quaity. Use of MPEG-4 coud therefore compensate for the higher bitrate demand of fat-pane dispays. A number of countries, where DVB-T is yet to start, are considering or have aready decided to use MPEG-4 with DVB-T. However countries that have aready introduced DVB-T wi need to use additiona ayers for introducing MPEG-4, with the exception of France where some mutipexes have aready been introduced with MPEG-4 (for pay-tv). Currenty DVB-H piot transmissions are taking pace in severa countries and are aready operationa in Itay [15]. In panning DVB-H services, a baance needs to found between, on the one hand, the radiated power and number of transmitters required to obtain the wanted coverage and, on the other hand, the avaiabe bitrate. As the reception conditions are very demanding [8], most operators tend to choose a robust system variant with the consequence of a imited net bitrate. Therefore 10 to 15 programmes may be accommodated in a DVB-H mutipex. It is expected there wi be more than 50 miion HD-ready TV sets in Europe by 2010 [16] and, consequenty, there wi aso be a high demand for HDTV programmes. Currenty HDTV programmes are deivered by sateite but many European broadcasters are panning to transmit HDTV on terrestria networks. EBU studies [17] indicate that two HDTV programmes can be accommodated in a DVB-T mutipex for rooftop reception (24 Mbit/s). HDTV is not compatibe with standard definition TV reception and therefore HDTV 6.2. UMTS considerations One option for use of the digita dividend, which is being considered by CEPT, is UMTS. The UMTS Forum considers that 2 x 30 MHz of paired spectrum, based on 5 MHz channeing, woud provide a viabe minimum coverage extension band for UMTS [18]. This requirement incudes a guard band between the upink and downink sub-bands and woud aso require guard bands of 10 to 16 MHz between it and the adjacent sub-bands used for DVB-T Other uses Assignments to other services having primary status in the Radio Reguations have been taken into account at RRC-06 if so requested by the Administrations concerned. These services incude radio navigation and fixed or mobie services for miitary appications and are shown in the List of Annex 5 of GE06. In any re-panning process, if so required, these services need to be taken into account. In addition there are services with secondary status in the Radio Reguations in Band IV/V. These services are not taken into account when primary services are panned. However, on a nationa basis, these services coud be of great importance, for instance the Radio Astronomy Service in channe 38 and Services Anciary to Broadcasting and Programme making (SAB/SAP). SAB/SAP services are of increasing importance because an increase in the number of broadcast programmes means aso an increase in the need for faciities to produce broadcast programmes. This is true in spite of the fact that the use of SAB/SAP in Band IV/V is becoming more restricted since the band is densey panned for DVB-T, eaving ess room for SAB/SAP transmissions EBU TECNICAL REVIEW 9

10 SPECTRUM MANAGEMENT 7. Digita Dividend choices From a technica point of view there are two aternative options for digita dividend appications: Either, Appications making use of Pan entries that require no or imited modifications to the GE06 Agreement, such as DVB-T, HDTV, DVB-H Some restrictions may be expected because of power imitations and interference eves of the corresponding GE-06 Pan entries. In most cases the services can be impemented under Artice 5 of the GE-06 Agreement and no internationa agreement is needed. In some cases pan modifications may be needed by appying Artice 4 of the GE-06 Agreement, requiring the agreement of potentiay affected countries. Adjacent channe probems may occur if different network topoogies in Band IV/V are used in the same area. These probems need to be soved nationay. Or, Appications making use of a dedicated sub-band with the consequence of considerabe modifications to the GE06 Pan. In case of upink transmissions, an aocation in the Radio Reguations for Mobie services woud be needed. In addition, guard bands are needed. For new appications, Artice 4 of the GE06 Agreement needs to be appied. Restrictions are to be expected in order to protect the GE06 Pan entries of other countries; interference from Pan entries of other countries needs to be accepted. Some technica constraints may arise if different network topoogies and systems co-exist in the same bands. Feasibiity studies are needed. Re-panning of the remaining part of the band is needed for DVB-T, requiring appication of Artice 4 of the GE06 Agreement. A transition from the origina GE06 pan to a modified pan is needed. Re-panning and transition to a modified pan wi be a compex and time-consuming process requiring severa years of intense internationa coordination. In severa European countries, five or six mutipexes for DVB-T or DVB-H have been icensed or wi be icensed soon. This means that in those countries a considerabe part of the digita dividend wi be used for categories 1 and 2 (see Section 1). After having icensed five or six mutipexes for DVB-T or DVB-H, in genera one or two ayers remain. These coud in principe be considered for a three digita dividend categories. For appication of the third category ony (a new use such as 3G), a dedicated subband must be considered and consequenty a re-panning process. However it raises the foowing questions: Woud WRC-07 or WRC-11 indicate Band IV/V as a 3G extension band when there are so many aternative bands, whie Bands III, IV and V are the ony possibiities for wide-area coverage of DVB-T and DVB-H? Woud Administrations be interested in invoving themseves in another intensive period of re-panning for digita broadcasting with unpredictabe resuts after having experienced the two sessions of the RRC and more than six years of preparing for these? Woud broadcasters and network operators be wiing to bear the nuisance and the costs of another transition period without the benefit of transmitting additiona services? Ony the future wi te! Jan Doeven received a bacheor degree in Eectrica Engineering in A through his career, he hed eading positions in frequency management and the appication of new technoogies for broadcasting. He worked for Nozema and KPN Broadcast Services in the Netherands as Strategic Technoogy Advisor unti his retirement in August 2007 and he is now an independent consutant. He has participated in EBU activities in the fied of radio and teevision broadcasting for 30 years and was chairman of the Broadcast-technoogy Management Committee (BMC) from 1997 to Since the eary nineties, Jan Doeven has been deepy invoved, nationay and internationay, in the panning and impementation of digita broadcasting networks. He chaired the European preparatory groups for RRC-04 and RRC-06 (CEPT Project Team FM24 and CEPT Working Group RRC-06 respectivey) and during RRC-06 he was the overa CEPT coordinator and vice chairman of the Conference. 10 EBU TECNICAL REVIEW 2007

11 SPECTRUM MANAGEMENT References [1] Fina Acts of the RRC-06 and associated Frequency Pans and List ITU, Geneva, 16 June 2006 [2] Draft RSPG Opinion on EU spectrum poicy impications of the digita dividend [3] Communication from the Commission to the Counci, the European Pariament, the European Economic and Socia Committee and the Committee of the Regions; EU spectrum poicy priorities for the digita switchover in the context of the upcoming ITU Regiona Radiocommunication Conference 2006 (RRC-06) COM(2005) 461, Brusses [4] Digita dividend Editoria by Phiip Laven, EBU Technica Review No. 308 (October 2006) [5] Fina Acts of the Regiona Radiocommunication Conference for the revision of the Stockhom 1961 Agreement (RRC-06-Rev. ST61) Geneva, 16 June 2006 [6] EBU doc. Tech 3313: Band I Issues EBU, August 2005 [7] Communication from the Commission to the Counci, the European Pariament, the European Economic and Socia Committee and the Committee of the Regions; on acceerating the transition from anaogue to digita broadcasting COM(2005) 204 fina, Brusses, [8] Terry O Leary, Eena Puigrefagut and Waid Sami: GE06 overview of the second session (RRC-06) and the main features for broadcasters EBU Technica Review No 308, October 2006 [9] EBU doc. Tech 3317: Panning parameters for hand-hed reception, considering the use of DVB-H and T-DMB in Bands III, IV, V and 1.5 GHz EBU, November 2006 [10] Radio Spectrum Poicy Group Opinion on the introduction of mutimedia services in particuar in the frequency bands aocated to the broadcasting services EU, 25 October 2006 [11] Draft CPM report, Chapter 1, ITU-R Document CPM07-2/1-E; 4 October 2006 [12] EBU doc I : Guideines for the RRC-06 EBU, Geneva [13] EBU doc I : The potentia impact of fat pane dispays on broadcast deivery of teevision EBU, Geneva [14] EBU doc I : Maximizing the quaity of conventiona quaity broadcasting in the fat pane environment EBU, Geneva [15] DVB-H services; dvb-h.org/services.htm [16] High Definition Teevision: Goba Uptake and Assessment To 2010 Screen Digest, March 2006 [17] EBU doc Tech 3312: Digita Terrestria HDTV Broadcasting in Europe; The data rate capacity needed (and avaiabe) for HDTV EBU, February 2006 [18] Coverage Extension Bands Report No. 38 from the UMTS Forum EBU First pubished : January EBU TECNICAL REVIEW 11

12 DIGITAL AUDIO BROADCASTING The evoution of DAB Frank Herrmann Chairman, WordDMB Technica Committee Larissa Anna Erismann Chairperson, WordDMB Marketing Committee Markus Prosch Chairman, WordDMB TC Task Force on Audio Systems DAB aready covering 500 miion peope in 40 countries around the word represents the fuy mobie and narrowband (1.7 MHz) terrestria branch of COFDM broadcasting technoogies. Athough the famiy of DAB standards has been growing continuousy from its beginnings in the eary 90s, severa major miestones have been reached by the WordDAB / WordDMB Forum, especiay within the ast three years. The most prominent exampes are certainy DMB and DAB+. Those two and further appications, as we as the necessary framework created, are iustrated in this artice. The technica perspective is accompanied by an economic one, visuaising the growth underway and the promising prospects that ie ahead, based on the substantiay extended DAB tookit. 1. Introduction 1.1. Fexibiity and Reiabiity keys to the success of Eureka-147 In the digita age, broadcasting technica standards need to baance the benefits of stabiity and innovation. Stabiity gives confidence to broadcasters, manufacturers and consumers. And yet, enhancing a standard to take advantage of technoogica innovation can offer new benefits and protect a standard s competitiveness in a rapidy changing market pace. The internationa organization responsibe for the Eureka-147 standards, WordDMB (formery known as the WordDAB Forum) has carefuy monitored deveopments in audio and mutimedia broadcasting over the ast decade and has kept up to date with state-of-the-art coding and transport systems. Athough WordDMB remains a strong advocate of stabiity, innovation in the interest of efficiency and diversity is an important issue in today s highy competitive market, and the Eureka-147 famiy of standards has easiy managed to keep on top of the ever-increasing speed of deveopments in the digita word. Severa chaenges have been met over the years and overcome by the addition of new features to the Eureka-147 famiy, each time increasing its fexibiity even further whist ensuring a continuing robustness and reiabiity. 12 EBU TECNICAL REVIEW 2007

13 DIGITAL AUDIO BROADCASTING When the origina DAB (Digita Audio Broadcasting) system was first deveoped in the ate 1980s, it was based on MPEG Audio Layer II coding, which was then state-of-the-art and is sti a commony used coding technoogy in digita radio broadcasting. Since then, MPEG Audio Layer III, better known as mp3 has conquered the market of digita music payers and radio streams. Even though sti the most successfu technoogy on the market, mp3 has aready been overtaken in efficiency and performance by MPEG-4 AAC (Advanced Audio Coding). This deveopment has caed for an additiona audio coding system in DAB which woud aow for more efficiency at ower bitrates hence the birth of DAB +. Another important innovation has been the addition of video/mutimedia capabiities to DAB, aowing it to become a digita mobie teevision patform caed DMB (Digita Mutimedia Broadcasting) as we as a digita radio patform. Both for DMB and DAB+, the technica basis remains with DAB. In other words, the physica ayer is sti the same... just new appications, new transport protocos and a second error-contro coding ayer have been added (Fig. 1). New chaenges wi continue to be addressed by the WordDMB Technica Committee, which wi ensure DAB remains a very attractive, fexibe and market-ready standard for digita audio, mobie and mutimedia broadcasting. One of the strengths of the Eureka-147 DAB standards is that not ony different appications can co-exist within the same mutipex, but aso different transport protocos and individua convoutiona code rates for each subchanne respectivey Tested, triaed and roed out a over the word Thanks to this fexibiity and robustness, Eureka-147 standards have managed to conquer more than 40 countries a over the word (Fig. 2). DAB digita radio, for exampe, has been tested, triaed and roed out in most European countries, among them significant markets such as the UK, Germany, Spain, Itay, France, Denmark, Norway and Switzerand. Digita Audio Broadcasting has traveed further overseas to numerous countries in Asia, Africa and America, and has aso arrived in Austraia and New Zeaand. Figure 2 The word of Eureka-147 Mobie teevision using DMB, on the other hand, was successfuy introduced in Korea in November 2005 and has since become the biggest mobie teevision market in the word. Ony 18 months after its aunch, the DMB receiver market in Korea passed the 4-miion mark at the end of March 2007 (Fig. 3). In Germany (DMB) and the UK (DAB- IP), mobie TV is currenty in its initia stages of ro-out, and there have been numerous tests and trias on DMB in other countries, incuding France, China, Norway, Denmark, India, Germany and the UK. Figure 1 DAB System Protoco Stack Aso, the additiona audio codec used by DAB+, even though it has ony just been issued as a standard, has aready raised much interest, especiay in markets where digita radio is about to be first roed out. The two most prominent DAB+ bidders are Austraia, where the introduction is panned to start in 2008/9 in the 11 most popuated cities, and Mata where icences for digita radio using 2007 EBU TECNICAL REVIEW 13

14 DIGITAL AUDIO BROADCASTING DAB+ have recenty been acquired by DigiB, a network operator. Many other countries are panning trias and tests for DAB+, among them Switzerand, Itay, Luxembourg, Begium, France, India and South Africa DAB/DMB receiver market on a steady rise It is not surprising, therefore, that the DAB/DMB receiver market has deveoped rapidy over the ast five years. Apart from the four miion DMB receivers that have been sod in Korea since the commercia aunch of T-DMB in November 2005 (Fig. 3), there are aso more than five miion DAB radios now in European househods, most of them in the UK (Fig. 4). DAB/DMB receivers are avaiabe in a price ranges, and the atest figures (Fig. 5) show that the choice of different receiver modes is simpy staggering. Over 250 manufacturers offer a tota of amost 900 different receiver modes, and the end of this growth is not yet in sight the offer is now amost twice as arge as it was just over a year ago. 2. Audio: the origina DAB and the additiona DAB DAB During the deveopment of DAB in the eary nineties, MPEG-1 (samping rate = 48 khz) and MPEG-2 (samping rate = 24 khz) Layer II audio coding were seected as the most appropriate agorithms at that time. MP3 (i.e. Layer III) was refused, because a better performance coud not be verified at that time, but higher processing power was required. Layer II was very robust against the errors imposed by the broadcast channe and was protected we enough by the convoutiona channe coding and the time intereaving introduced as part of the physica ayer of the OFDM-based broadcasting system DAB DAB+ After many years of repeated discussions and consideration, the point of time was reached in 2005 for making use of the remarkabe margin that audio compression deveopments had ed to within a decade. Pressure in this direction was generated especiay by markets about to start with the ro-out of DAB. Naturay they were not ready to ignore the deveopments sketched above. In addition, existing DAB markets were aso ooking for expansion through efficiency enhancements. Another eement of the arising pressure was the fact that some providers were considering or had aready started to make use of the appication DMB for carrying just audio services. For cassica radio, DMB with its state-of-the art audio codecs HE-AAC v2 and MPEG-4 ER BSAC ooks attractive on first view, but audio services can be reaised in a much more efficient and smarter way. So the WordDAB Forum decided in June 2005 to start the deveopment of an aternative audio system for DAB the / / / /2007 Figure 3 DAB receiver saes in the UK (accumuative) W 27 / 2004 W 37 / 2004 W 47 / 2004 W 5 / 2005 W 15 / 2005 W 25 / 2005 W 35 / 2005 W 45 / 2005 W 3 / 2006 W 13 / 2006 W 23 / 2006 W 33 / 2006 W 43 / 2006 W 43 / 2006 W 01 / 2007 W 11 / 2007 W 21 / '000'000 5'000'000 4'000'000 3'000'000 2'000'000 1'000' W 27 / Figure 4 Saes of DMB receivers in Korea W 37 / 2004 W 47 / 2004 W 5 / 2005 W 15 / 2005 W 25 / 2005 W 35 / 2005 W 45 / 2005 Dec 2002 Feb 2003 Apr 2003 Jun 2003 Aug Dec 2003 Feb 2004 Apr 2004 Jun 2004 Aug 2004 Oct 2004 Dec 2004 Feb 2005 Apr 2005 Jun 2005 Aug 2005 Oct 2005 Dec 2005 Feb 2006 Apr 2006 Jun 2006 Aug 2006 Oct 2006 Dec 2006 Feb 2007 Apr 2007 W 3 / 2006 W 13 / 2006 W 23 / 2006 W 33 / 2006 W 43 / 2006 W 01 / 2007 W 11 / 2007 W 21 / 2007 Figure 5a Manufacturers of DAB/DMB receivers (Source: May 2007) Figure 5b Numbers of DAB/DMB receivers modes (Source: May 2007) 14 EBU TECNICAL REVIEW 2007

15 DIGITAL AUDIO BROADCASTING Technica Committee set up the Task Force, New Audio System. The resut of 1.5 years of enthusiastic work the norm Transport of Advanced Audio Coding (AAC) audio was pubished by ETSI in February 2007 and was announced pubicy as DAB+ at the same time. The significanty increased efficiency, which is discussed in more detai ater, offers benefits for Governments and Reguators (even better spectrum efficiency), broadcasters (ower costs per radio station) and consumers (a bigger choice of stations). It is designed to provide the same functionaity as the current MPEG Audio Layer II radio services. sef-expaining to decide in favour of the most recent deveopment HE-AAC v2. It sti enabes the appication of, for exampe, just the core codec for high fideity radio at the higher bitrates. Providers have the choice of using just the core, the core pus SBR... or the core pus SBR pus PS. Of course, the receivers must be prepared for a cases and hence the impementation of HE-AAC v2 is mandatory. In ight of the fact that audio coded with MPEG Layer II wi remain on-air for many years to come, a new DAB receiver needs to cover both coding agorithms MPEG-1/2 Layer II and HE-AAC v2. DAB was originay designed around MPEG-1 ayer II structures best refected by the fact that the DAB ogica frames were of identica ength in time (24 ms) as the MPEG-1 ayer II audio frames. The step to MPEG-2 Layer II, with haf the samping rate, was simpe one audio frame per two ogica frames. And DAB+ uses the common denominator of a permitted engths of AAC Access Units (of ength 20, 30, 40 or 60 ms) with a In some countries where DAB digita radio has aready been aunched, broadcasters are committed to continuing to use MPEG Audio Layer II. However, in countries panning to aunch digita radio, the arguments in favour of aunching DAB+ are compeing. It is worth noting that this is not the first time HE-AAC v2 has been incuded in the Eureka-147 famiy of standards. Aready, the DMB standard aows for HE-AAC v2 audio as part of the video services. However, DMB designed for mobie teevision naturay acks some of the functionaity required for pure radio services. Other broadcast technoogies such as DVB-H (digita video broadcasting to handhed devices), DRM (Digita Radio Mondiae; i.e. digita ong-, medium- and short-wave) or Quacomm s MediaFLO technoogy aso use HE-AAC v2 audio coding and are abe to carry mutipe audio services in the digita capacity needed for a singe radio station using MPEG Audio Layer II Technica overview of DAB+ The corresponding Ca for Technoogies resuted in just one famiy of audio codecs desired by the group of appicants AAC. Since AAC is buit up as a hierarchica system (see Fig. 6 and the text box), it was HE-AAC v2 DAB+ uses MPEG-4 High Efficiency AAC v2 profie (HE-AAC v2). This audio codec is the most efficient audio compression scheme avaiabe wordwide. It combines three technoogies: The core audio codec AAC (Advanced Audio Coding). A bandwidth extension too SBR (Spectra Band Repication), which enhances efficiency by using most of the avaiabe bitrate for the ower frequencies (ow band) of the audio signa. The decoder generates the higher frequencies (high band) by anaysing the ow band and side information provided by the encoder. This side information needs consideraby ess bitrate than woud be required to encode the high band with the core audio codec. Parametric Stereo (PS): a mono down-mix and side information is encoded as opposed to a conventiona stereo signa. The decoder reconstructs the stereo signa from the mono signa using the side information. HE-AAC v2 is a superset of the AAC core codec. This superset structure permits the use of (i) pain AAC for high bitrates, (ii) AAC and SBR (HE-AAC) for medium bitrates or (iii) AAC, SBR and PS (HE-AAC v2) for ow bitrates. Therefore HE-AAC v2 provides the highest eve of fexibiity for the broadcaster. A detaied description of HE-AAC v2 is avaiabe on the EBU website 1. An introduction to MPEG-4 is avaiabe on the MPEG Industry Forum website 2. HE-AAC v2 provides the same perceived audio quaity at about one third of the subchanne bitrate needed by MPEG Audio Layer II. The same audio coding is aso used in DRM and DMB e.g. for teevision audio. Devices, which aso incude DMB or DRM can benefit from the fact that the audio coding for this range of technoogies is essentiay the same. 1. EBU Technica Review: MPEG-4 HE-AAC v2 audio coding for today s digita media word (2006) 2. An MPEGIF White Paper: Understanding MPEG-4: Technoogies, Advantages, and Markets EBU TECNICAL REVIEW 15

16 DIGITAL AUDIO BROADCASTING 120 ms ong superframe equivaent to five ogica frames (Fig. 7). It shoud be noted here that this quite short ength, and hence the quick zapping from one service to another, was reaised through the adoption of the AAC variant with 960 sampes per Access Unit (as used for Digita Radio Mondiae). PAD mutimedia (Dynamic Labes such as tite, artist information or news headines, sti images such as weather charts, and other mutimedia content); service anguage and programme type information (e.g. Cassica Music, Rock Music, Sport); etc.. Due to the high efficiency of the new coding agorithms, the impact of ost bits is more significant. In other words, better protection is needed. Aready introduced for DMB more precisey for DAB Enhanced Stream and Packet Mode the concatenation of the inner convoutiona coding (Viterbi), being an eement of the origina DAB set-up, and an outer bock code in the form of Reed-Soomon (R-S) coding was chosen as the most appropriate soution. The structure appied (Fig. 8) consists of super-frames covering a fixed number of AAC access units. Each Access Unit (AU) carries its PAD (Programme Associated Data) part in a simiar way as for MPEG Layer II audio frames. The required additiona error protection is reaised with virtua intereaving and an R-S scheme (120, 110, t=5) derived from the same mother code as the R-S schemes for Enhanced Stream and Packet Mode. The ten parity bytes per 110 data bytes equivaent to an overhead of 8.3% ead to an abiity of correcting up to five erroneous bytes in those 120 bytes (Fig. 9). For test purposes, the new agorithms have aready been impemented in both transmitting and receiving equipment. Therefore the step towards mass production is a sma one for those who have aready invested effort and resources in the standardization exercise. Figure 7 DAB Logica Frame Aignement for Layer II and DAB+ Figure 8 Superframe structure used for transport of HE AAC v2 audio in DAB 120 coumns * 4 rows Features of DAB+ A the functionaity avaiabe for MPEG Audio Layer II services is aso avaiabe for DAB+: service foowing (e.g. to FM or other DAB ensembes); traffic announcements; 110 audio/pad/man FEC coumns Figure 9 Error-contro code cacuation and virtua intereaving in a 32kBit/s sub-channe 16 EBU TECNICAL REVIEW 2007

17 DIGITAL AUDIO BROADCASTING Abbreviations AAC ASF AU BER BIFS BSAC C/N CRC CTS DAB DAB+ DAB-IP DLS DMB DRM DRM DVB EPG EPM ETSI FEC FM GE06 Advanced Audio Coding (Microsoft) Advanced Streaming Format Access Units Bit-Error Ratio BInary Format for Scene description Bit Siced Arithmetic Coding Carrier-to-Noise ratio Cycic Redundancy Check (MPEG) Composition Time Stamp Digita Audio Broadcasting (Eureka-147) org/ DAB using the AAC codec DAB Internet Protoco Dynamic Labe Segment Digita Mutimedia Broadcasting Digita Radio Mondiae Digita Rights Management Digita Video Broadcasting Eectronic Programme Guide Enhanced Packet Mode European Teecommunication Standards Institute queryform.asp Forward Error Correction Frequency Moduation Geneva Frequency Pan of 2006 The mutimedia information carried in the PAD of an HE-AAC v2 radio service is as we protected against data osses as the audio itsef, both enjoying the cascaded error-contro coding. In order to ensure that the PAD data of a radio service using MPEG Audio Layer II aso takes advantage of the new deveopments, a backwardscompatibe and optiona FEC ayer wi be added here as we. HE-AAC IOD IP MOT MPEG MSC OCR OD OFDM PAD PCR PES PID PMT PS PSI R-S RDS SBR SD SI SL TS UDP WMA XML High Efficiency AAC Initia Object Descriptor Internet Protoco Mutimedia Object Transfer Moving Picture Experts Group chiarigione.org/mpeg/ Main Service Channe Object Cock Reference Object Description Orthogona Frequency Division Mutipex Programme-Associated Data Programme Cock Reference Packetized Eementary Stream Packet IDentification number Programme Map Tabe Parametric Stereo Programme Service Information Reed-Soomon Radio Data System Spectra Band Repication Secure Digita Service Information Synchronization Layer Transport Stream User Datagram Protoco (Microsoft) Windows Media Audio Extensibe Markup Language An important design criterion for DAB+ was a short zapping deay. Both the time it takes to switch from one radio station to another station on the same DAB ensembe, as we as the time it takes to tune to a radio station on another DAB ensembe, was minimized. Currenty a DAB radio stations are mono or stereo. However, DAB+ aso provides the means to broadcast surround sound in a backwards compatibe way. Using MPEG Surround it is possibe to broadcast a stereo signa together with surround side information (e.g. 5 kbit/s side information). Standard stereo radios wi ignore this side information and decode just the stereo signa. MPEG Surround receivers wi evauate the side information and reproduce surround sound. So at a comparativey ow additiona bitrate, the broadcaster can increase the audio experience on surround sound receivers, and sti provide high quaity sound to a other radios Performance of DAB+ During the standardization process, simuations were undertaken by the Communications Research Centre, Canada. With typica reference channe modes for DAB environments, a gain of 1.7 to 6.7 db for the new system, compared to the existing Layer II system, was determined. Fied tests conducted in the UK and Austraia confirmed the resuts of the simuations. They showed that the geographica coverage area of radio services using HE-AAC v2 is sighty arger than that for radio services using MPEG Audio Layer II. Audio services using HE-AAC v2 performed about 2-3 db better at the threshod of audibiity. This means that in some areas cose to the coverage area imit, where MPEG Audio Layer II services aready showed audibe artefacts, HE-AAC v2 radio services showed no audibe artefacts. The error behaviour of MPEG Audio Layer II is different to that of HE-AAC v2. With MPEG Audio Layer II, the weaker the DAB signa gets, the more audibe artefacts can be heard. HE-AAC v2 produces no audibe artefacts, but when the signa gets too weak, an increased number of audio frames wi be ost and this causes short periods of sience (fade-out and fade-in). Test isteners preferred this error behaviour EBU TECNICAL REVIEW 17

18 DIGITAL AUDIO BROADCASTING Figure 10 C/N for threshod of audibiity (Source: Communications Research Centre Canada) Compared to radio services using MPEG Audio Layer II, radio services using HE- AAC v2 wi fai ater (they can cope with a sighty ower DAB signa quaity), but the margin from error-free reception to oss of reception is smaer (Fig. 10) Impementation scenarios Thanks to the fexibe structure of the DAB system, radio services encoded with MPEG Layer II can co-exist with radio services encoded with HE-AAC v2. Exampes of mutipex impementations are given in Fig ) First on the eft: this is the cassica set-up with, say, nine MPEG Layer II encoded radio services. 2) Second from the eft: in contrast to the cassica set-up, a progressive consteation is shown here. It does not aow for egacy receivers not understanding the new coding agorithm and no fewer than 28 DAB+ radio services can find space in such a mutipex arrangement. 3) Second from the right: this is a migration scenario that moves sowy from Layer II to AAC. It is shown sti providing five Layer II services, but aready bringing 11 AAC-coded services on air. 4) First on the right: this is another way of benefitting from the saved Ensembe capacity. With three Layer II services and eight AAC services, there is sti enough capacity eft for two mobie TV services using DMB. 3. Mobie TV: DMB and BT Movio In eary 2007, there were two different variants for providing video appications via DAB DMB and DAB IP Tunneing. An exampe for the atter path was known in the UK as BT Movio and was provided as a whoesae service by the incumbent teecom operator. This appication put a source coding agorithm that was not specified with an open standard on top of IP. In fact, the whoe appication was a proprietary one. 3.1 Digita Mutimedia Broadcasting (DMB) DMB uses H.264/MPEG -4 AVC (Advanced Video Coding), HE- AAC v2 or BSAC (Bit-Siced Arithmetic Coding) and BIFS (Binary Format for Scenes), respectivey, as the encoders for video, audio and content-reated data services. A of these encoded Eementary Streams are mutipexed into MPEG-2 Transport Stream (TS) packets. To increase the necessary robustness especiay for mobie reception an additiona bock coding scheme (Reed- Soomon Coding) and convoutiona intereaving is appied to the MPEG-2 Transport Stream in ine with DVB structures. The byte-intereaved and error-protected TS packets are transmitted through the Eureka-147 stream mode. T-DMB obtained officia approva as a European ETSI standard in Juy Extraction, error-contro decoding, stripping of Eementary Streams and synchronization both temporay and spatiay as we as source decoding and reproduction are shown in Fig. 12 for the termina side. Atogether, this chain represents a cassica combination of MPEG-4 eements transported by an MPEG-2 Transport Stream. BIFS, as one of those MPEG eements/norms, represents quite a powerfu too for data provision and interactivity. 3.2 BT Movio Unike DMB, BT Movio was not fuy standardized by WordDMB / ETSI, but made use of a hook that was designed exacty for that purpose DAB IP Tunneing. Based on this transport system for IP datagrams via DAB, the provider appied protocos and source coding agorithms designed by Microsoft. It shoud be noted that, in the meantime, a of these specifications (ASF and VC-1) are in the pubic domain apart from one WMA. As with most DAB data formats, IP Tunneing is based on (Enhanced) Packet Mode see Figs 1 and 20. The encapsuation of the IP datagrams in DAB MSC Data Groups (DGs) either unfragmented or fragmented is shown in Fig. 13. Figure 11 Various impemantations of the DAB system For the unfragmented case, the size of a singe MSC Data Group data fied, carrying aways exacty one IP datagram, ies in the range 576 to 8191 bytes. This is given by the minimum size of an IP Datagram according to RFC 791 and the argest MSC Data Group size according to ETSI standard EN EBU TECNICAL REVIEW 2007

19 DIGITAL AUDIO BROADCASTING Dispay and User Interaction For the fragmented case, the MSC DGs might be even smaer. Mapping of the Data Groups onto Packets is done as usua, using arge packet sizes as far as possibe and imiting the padding both measures for reducing the overhead. Interactive Audiovisua Scene Composition and Rendering Especiay for streaming services such as BT Movio, it was absoutey vita to empoy the second ayer of error-contro coding. Compression Layer Sync Layer Object Descriptor Scene Description Information AV Objects Data SL SL SL SL SL SL Upstream Information On top of IP, BT Movio need UDP and ASF. Source materia was encoded with Windows Media Audio and Video codecs (the atter one is equivaent to VC-1). BT Movio services were a digita-rights managed according to a Microsoft DRM specification. They were enhanced with the DAB Eectronic Programme Guide. Deivery Layer MPEG-2 TS (PES) Outer Coder (RS + Intereaver) Stream Mode of ETSI EN The BT Movio device marketed in the UK was the HTC/Qtek Lobster (Fig. 14). Of course, this device was aso equipped for the reception and reproduction of DAB radio services. Figure 12 Termina processing chain for the DAB appication, DMB Some thes artice was first pubished the BT Movio service has cosed down. IP Datagram IP Header Data 4. Further appications MSC Data Group Data Group Header Data Group Data Fied CRC (optiona) 4.1.Inteitext Packets Packet Header IP Datagram arger than MTU Fragments of IP Datagram Packet Data Fied IP Header (fragment #1) IP Header CRC Data Packet Header Data IP Header (fragment #n) Packet Data Fied Data CRC Inteitext is the youngest offspring of the Eureka-147 famiy of standards. It extends the we-known Dynamic Labe in a backwards-compatibe and structured way and aows for the provision of text eements, enabing a hierarchy of detai. Inteitext is transported in PAD excusivey. MSC Data Groups Packets Packet Header Figure 13 DAB IP Tunneing Data Group Header Data Group Data Fied Packet Data Fied CRC CRC (optiona) Data Group Header Packet Header Data Group Data Fied Packet Data Fied CRC (optiona) CRC The data is compied into a simpe Tee/ Videotext-ike database of information which the user of any DAB radio, equipped with this appication, can browse on demand. Inteitext messages are a specia form of Dynamic Labe messages, formatted in such a way that receivers not supporting Inteitext wi continue to function normay EBU TECNICAL REVIEW 19

20 DIGITAL AUDIO BROADCASTING taksport Footba Prem Lge Tabe 1. Chesea 27 pts 2. Spurs 18 pts 3. Charton 16 pts 4. Man Utd 14 pts 5. Man City 14 pts Resuts Arsena 0, Wigan 3 Spurs 2, Man Utd 1 West Ham 2, Sunderand 3 Figure 14 HTC/Qtek Lobster, as used with the BT Movio DAB-IP service Figure 15 Exempe of dispay of Inteitext message data Figure 16 Receiver with Inteitext Capabiity: PURE digita One Since not a transmitted DL messages wi be Inteitext messages, Inteitext-capabe receivers need to determine whether a received DLS message is an Inteitext message or not, in order to process the received message appropriatey. Inteitext messages are parsed and stored. The stored messages are updated and deeted to ensure that the data is appropriatey maintained. The Inteitext system provides a means for broadcasters to contro the ifetime and basic formatting of broadcast information, whie the dispay of information is userdriven. The Inteitext system aows the broadcaster to dictate the structure and design of menus, incuding menu naming. The information provided by each service provider is stored in such a way that it cannot be atered by any other service provider. Navigation is usuay via a simpe up/ down/seect interface, with the actua dispay being taiored to the resources avaiabe to a given receiver. Inteitext messages consists of a category, a sub-category and some data. Within a given category, the sub-categories may be ordered by using a numerica index; simiary the data items are ordered within the sub-category. An exampe of the type of user dispay is shown in Fig. 15. Once again, the UK is the first market introducing the new technoogy: it wi be possibe for a number of existing receivers (Fig. 16) to be updated for reproducing this more attractive, but sti simpe to use, text appication EPG Who coud imagine teevision today in the absence of an EPG? And for radio it s even Service information ENSEMBLE SERVICE Schedue information SCHEDULE PROGRAMME EVENT Group information GROUP GROUP PROGRAMME Figure 17 EPG entities and hierarchies Information about an ensembe more important. The DAB Eectronic Programme Guide aso suitabe for Digita Radio Mondiae is avaiabe in two variants, binary and XML. It provides an overview of the Programme Items currenty on-air and the ones that wi be on-air within a given time period, e.g. over the next 24 hours. It is aso appicabe to Mobie TV services. The EPG might Information about a service and a ink to the ensembe it broadcasts on Information about a schedue for one or more services Information about a programme and a ink to its service Information about an event and a ink to its programme Information about a group Information about a group and inks to parent groups A programme inks to its parent groups 20 EBU TECNICAL REVIEW 2007

21 DIGITAL AUDIO BROADCASTING see the user interface at the top of the figure. 5. Second error-contro coding ayer Figure 18 EPG-enabed DAB Receivers cover the tuned Service, severa or a Services on the tuned Ensembe or it can even incude Services being broadcast on other Ensembes. With this technica too at hand, the next step pre-programmed Service seection and/or recording (nowadays on SD Cards) is the ogica way towards a state-of-theart receiver. EPG content can be generated without requiring dedicated production teams. Fig. 18 shows some consumer receivers that feature the appication EPG, e.g. in the form of dispaying programme schedues for the next seven days. Timed recording is enabed as we The more efficient state-of-the-art sourcecoding agorithms are naturay more sensitive to transmission errors. Here the origina agorithms were significanty more robust and error-toerant. H.264 video coding requires an average BER as ow as 10-8 at the input to the decoder (Fig. 19). In contrast, DAB was originay designed for a BER of 10-4 at the input to the MPEG Layer II coder. After a thorough simuation and fiedtest project, this issue was soved in the end through the appication of a second error-contro coding ayer resuting in a cascaded coding arrangement with convoutiona Viterbi coding as the inner coding and Reed-Soomon bock codes as the outer coding as in DVB. Here, with an overhead of 7.8%, a reay dramatic improvement coud be reached. E P G s a r e t r a n s p o r t e d w i t h t h e Mutimedia Object Transfer (MOT) protoco and might be compressed for broadcast efficiency purposes see Fig. 1 above. Fig. 17 iustrates the hierarchies of the different sorts of information that can be provided Service Information, Schedue Information and Group Information. On the provider side, EPG is aready in use to a wide extent and its coverage is getting arger continuousy. What are the advantages for the providers? EPG enhances the istening experience, and is a marketing too for Digita Radio stations: EPG enabes additiona media spend, revenue stream, opportunities; Figure 19 Error behaviour of DAB with cascaded error-contro coding 2007 EBU TECNICAL REVIEW 21

22 DIGITAL AUDIO BROADCASTING MPEG-2 Transport Stream Mutipexer Reed-Soomon Outer Coder Outer Intereaver DAB MSC Stream Data Sub-channe Figure 21 Adapting Transport Streams to existing DAB infrastructure Figure 20 Position of Enhanced Stream and Packet Modes in the DAB protoco stack Figure 22 Forney Intereaver for the transmission of Transport Streams via DAB 5.1. Enhanced Stream Mode (ESM) This Transport Mode an evoution of what is caed MSC Stream Data in the centra DAB standard, ETSI EN is in fact an additiona Packet Mode, consisting of a structure of 188-byte ong packets with 16 Reed-Soomon parity bytes attached. Furthermore, a Forney intereaver is appied to those FEC ed 204-byte ong packets. This structure is in use for DMB with the MPEG-2 Transport Stream see ETSI TS Originay this structure was introduced with a variants of the DVB system Enhanced Packet Mode (EPM) In a simiar way as for the Enhanced Stream Mode described above, the existing Packet Mode ( MSC Packet Data in ETSI EN ) was extended and improved with another ayer of errorcontro coding. In the case of EPM, virtua time intereaving is appied. Two figures might iustrate the improved structure. Fig. 23 presents the FEC frame that is fied verticay, packet by packet. Fuy fied with an integer number of packets, the Reed-Soomon parity bytes are cacuated horizontay over the 188 bytes in the same row. The same R-S scheme as for Enhanced Stream Mode ed to a simiar performance and was hence reused. A appication data coumns are read out verticay as they are fied and are transmitted foowed by the R-S parity bytes aso read out verticay. Fig. 24 iustrates the simpe set-up of the equipment on the transmitter side. Legacy receivers ignore the FEC packets, because they are different from ordinary data packets in two ways: a) FEC packets carry a Packet Address that doesn t correspond to a Service Component, and; b) they are of a different structure, incuding the position of the CRC. row 1 coumn Appication Data Tabe (2 256 bytes) RS Data Tabe (192 bytes) data packets generate FEC frame containing data packets and RS data encapsuate RS data into FEC packets FEC packets data packets subchanne data 12 Figure 23 Buiding the FEC Frame Figure 24 Adapting EPM to the existing DAB infrastructure 22 EBU TECNICAL REVIEW 2007

23 DIGITAL AUDIO BROADCASTING The Enhanced Packet Mode can be used for a appication data defined for Packet Mode, because EPM is a fuy backwardscompatibe extension and sha therefore be appied to reguar transmissions without exceptions. 6. MPEG-2/MPEG-4 system and FEC overhead demystified Every broadcasting system requires the addition of a particuar overhead on top of the main content to be transported. Typica exampes for such an overhead are synchronization signaing, error-contro coding as we as service-parameter signaing and metadata. In particuar, regarding the specific case of transporting narrowband appications with DMB, the figures given range from a few percent to haf of a stream. In order to give such discussion a reiabe scientific basis, an exampe case is discussed here in detai a streaming appication. Let s grab a time sice from a narrowband DAB sub-channe used for DMB, consisting of a singe stream. Because it s a common denominator of the entities we want to discuss, et it be seven seconds ong. The sub-channe bitrate for this exampe (other exampes can be derived from this exercise, ideay eased by a few ines of source code) is 40 kbit/s. Here k for kio is equivaent to An MPEG-2 Transport Stream fis the sub-channe competey, which means that within the seven seconds MPEG-2 TS packets a with 16 Reed-Soomon parity bytes attached can be transported. Starting with the PSI/SI signaing required for this case, we need to transport the two tabes PAT (Programme Association Tabe) and PMT (Programme Mapping Tabe) every 500 ms. Hence within seven seconds, 28 of those tabes occur and TS packets are eft for other purposes assuming that each of the tabes can be accommodated by one singe TS packet. The MPEG-4 Initia Object Descriptor (IOD) is part of the PMT. For a compete description of the object transported i.e. the appication data stream the MPEG-4 system ayer entity OD (Object Description) gets its PID and corresponding TS packets that are repeated every 500 ms. So 14 packets are assigned to OD and the remaining amount sums up to TS packets. These packets offer 184 bytes each for the payoad. So atogether 23, bytes are avaiabe for the payoad to be transmitted. For controing the 27 MHz receiver cock accuratey, the Programme Cock Reference PCR parameter is required every 100 ms. PCR is provided within the so-caed adaptation fied being part of a TS packet and ocated right after the 4-byte header. The PCR carries the same PID as the accompanied stream and can herewith be transported in the TS packets carrying the payoad. PCR traves in the Adaptation fied and occupies eight bytes per occurrence. In tota, 560 bytes wi be consumed by PCR in seven seconds. With this, 23, bytes are eft. MPEG-4 Access Units (AU) carry the MPEG-4-encoded content. Each AU is embedded in a Synchronization Layer (SL) packet and the SL packet in a PES packet. Insertion of PES packets into TS packets can be done in a fragmented way. Assuming the extreme case of a ength in time of 60 ms for each MPEG-4 AU, of them need to be transported within seven seconds. This is equivaent to the number of SL and PES packets empoyed for the transport. The PES packet overhead is five bytes per packet and the SL packet overhead is one byte. Hence the compete overhead for seven seconds is 700 bytes. Overhead [%] Figure 25 MPEG-2/MPEG-4 system ayer and FEC overhead Due to the fact that, every 700 ms, the Object Cock Reference (OCR) for synchronization of MPEG-4 objects and the Composition Time Stamp (CTS) each of them 33 bits ong are repeated, every eeventh PES/SL packet additionay carries 66 bits (nine bytes) of this overhead, which sums up to bytes within seven seconds. Subtracting these bytes from the 23, above, there are 22, bytes avaiabe for the transport of naked Access Units. This vaue can be converted to a bitrate of kbit/s remaining for the Access Units. It is equivaent to 64.24% of the sub-channe bitrate of 40 kbit/s. So the overhead for the exampe discussed is 35.76%. With this cacuation and the reated assumptions appied to severa more subchanne bitrates we get : Sub-channe bitrate [kbit/s] Overhead [%] Sub-channe bitrate[kbit/s] 2007 EBU TECNICAL REVIEW 23

24 DIGITAL AUDIO BROADCASTING Frank Herrmann born 1963 received an engineering dipoma from the Technica University of Braunschweig in Germany. Having ed Carion into the Eureka-147 DAB project, digita broadcasting became the core eement of his professiona career. At Panasonic, he initiay managed the DAB piot project Hessen. Within the EU-sponsored TPEG project, he chaired the Industry Group and buit the basis for commercia services now reaised within the mobie.info project. Today Mr Herrmann is Project Leader for DVB-T2 and DAB/DMB at Panasonic s Frankfurt abs and head of the Digita Broadcasting Team. Within the DVB-T2 group, he recenty accepted responsibiity for forming a team to dea with the design of the system ayer. Frank Herrmann beongs to the odest members of WordDMB and became TC Chairman in He has put emphasis on a thorough refurbishment of this broadcasting system and co-initiaised WordDMB s contacts to externa Fora with priority today on forging inks with the DVB Project. Larissa Anna Erismann works for Swiss Sateite Radio, a branch of the Swiss pubic broadcaster (SRG SSR idée suisse) which produces three ad-free, music-ony radio stations: Radio Swiss Cassic, Radio Swiss Jazz and Radio Swiss Pop. These three stations form an important part of the DAB-ony offer in Switzerand. Ms Erismann took charge of the B2B and B2C aspects of DAB marketing for a three stations in 2004 and subsequenty aunched the ony web patform to offer a comprehensive range of business and consumer information in four anguages on the DAB situation in Switzerand. Larissa Anna Erismann is an acknowedged expert on Switzerand s DAB market and is continuousy expanding her famiiarity with DAB in other countries. In Juy 2006, she acceded to the chair of the WordDAB Marketing Committee after participating in a number of the Committee s projects and heading a task force dedicated to the promotion of DAB in Germany. Markus Prosch studied computer science at the University of Erangen and gained his Dip.- Ing. degree in In 1995 he joined the Fraunhofer Institute for Integrated Circuits IIS. Since 1996, he has been deepy invoved in the specification of DAB standards. He is responsibe for the deveopment of the Fraunhofer MMDS (Mutimedia DataServer), a system designed to insert a kinds of data services onto DAB. Starting in 2001, Mr Prosch has aso been invoved in the specification of Digita Radio Mondiae (DRM) standards. He is responsibe for the deveopment of the Fraunhofer DRM-CS (DRM ContentServer), Fraunhofer s soution for the DRM broadcast chain. He now represents Fraunhofer IIS within WordDMB. In 2006, he was the Chairman of the WordDMB task force on audio coding. In summary, it is recognized that for appications requiring ow sub-channe bitrates, the combination of MPEG-2 and MPEG-4 system ayers eads to quite a significant overhead. For higher data rates the overhead is ess significant. 7. Outook DAB has been refurbished in a way that assures its further success in the near future. The appication of new coding agorithms has been enabed through a second ayer of error-contro coding in a way that is widey impemented for DVB aready. Existing structures can easiy co-exist with the new ones iustrated above. Further to the eements described here, the new DAB middeware approach as we as voice-reated appications are awaiting their competion shorty. 24 EBU TECNICAL REVIEW 2007

25 DIGITAL AUDIO BROADCASTING Furthermore, the PAD content of the MPEG-1/2 Layer II radio services sha be protected more thoroughy. Once again, the we-estabished R-S scheme wi be appied (same mother code). As far as text appications are concerned, the DAB equivaent to RDS RadioText +, i.e. Dynamic Labe +, wi be adopted shorty. From the industry s point of view, the different digita broadcasting systems sha be aigned more cosey in the future. Corresponding convergence activities are about to be started and wi buid a core subject for the next few years in digita broadcasting. Making IP the universa and common ayer for more or ess a communication systems, a sides can make use of existing appications and there is ess and ess need for deveoping bearer-specific upper-ayer eements. In 2007, it is aso time to consider backwards-compatibe amendments to the physica ayer of DAB in order to extend DAB s spectra efficiency. Competition between broadcasting systems over the coming decade wi put emphasis on this aspect. Ceary, such a step does not come for free, but wi require higher C/N ratios. This means, first of a, that conformity with GE06 panning parameters must be assured for each such step. In addition, European reguations wi not aow for high fied strengths. So it is necessary to detect the borderine and move coser to it for sure, the necessary investments wi pay off after quite a short time period. DAB wi buid further on its strengths fexibiity and reiabiity. Reated standards ETSI EN V1.4.1 ( ): Radio Broadcasting Systems; Digita Audio Broadcasting (DAB) to mobie, portabe and fixed receivers. ETSI EN V2.1.1 ( ): Digita Audio Broadcasting (DAB); Mutimedia Object Transfer (MOT) Protoco. ETSI ES V1.1.1 ( ): Digita Audio Broadcasting (DAB); Internet Protoco (IP) datagram tunneing. ETSI TS V2.1.1 ( ): Digita Audio Broadcasting (DAB); Broadcast website; Part 1: User appication specification. ETSI TS V1.1.1 ( ): Digita Audio Broadcasting (DAB); Broadcast website; Part 2: Basic profie specification. ETSI TS V2.1.1 ( ): Digita Audio Broadcasting (DAB); Broadcast website; Part 3: TopNews basic profie specification. ETSI TS V2.1.1 ( ): Digita Audio Broadcasting (DAB); MOT Side Show; User Appication Specification. ETSI TS : V1.3.1 ( ): Digita Audio Broadcasting (DAB); Registered Tabes. ETSI TS V1.2.1 ( ): Digita Audio Broadcasting (DAB); Data Broadcasting Transparent Data Channe (TDC). ETSI TS V1.1.1 ( ): Digita Audio Broadcasting (DAB); A Virtua Machine for DAB: DAB Java Specification. ETSI TS V1.1.1 ( ): Digita Audio Broadcasting (DAB); DAB-TMC (Traffic Message Channe). ETSI TS V1.2.1 ( ): Digita Audio Broadcasting (DAB); Digita Radio Mondiae (DRM); Transportation and Binary Encoding Specification for Eectronic Programme Guide (EPG) ETSI TS V1.1.1 ( ): Digita Audio Broadcasting (DAB); Data Broadcasting MPEG-2 TS Streaming. ETSI TS V1.1.1 ( ): Digita Audio Broadcasting (DAB); DMB video service; User Appication Specification. ETSI TS V1.1.1 ( ): Digita Audio Broadcasting (DAB); Transport of AAC audio. ETSI TS V1.3.1 ( ): Digita Audio Broadcasting (DAB); Digita Radio Mondiae (DRM); XML Specification for DAB Eectronic Programme Guide (EPG). ETSI TS 102 iii V1.1.1 (2007-ii): Digita Audio Broadcasting (DAB); Inteitext. ETSI TR V1.1.1 ( ): Digita Audio Broadcasting (DAB); Guideines and rues for impementation and operation; Part 1: System outine. ETSI TR V1.1.2 ( ): Digita Audio Broadcasting (DAB); Guideines and rues for impementation and operation; Part 2: Syste m features. ETSI TR V1.1.2 ( ): Digita Audio Broadcasting (DAB); Guideines and rues for impementation and operation; Part 3: Broadcast network. Appendix A: WordDMB poicy statement on DAB/DAB+/ DMB The WordDMB poicy statement reating to how the standards for DAB/DAB+ and DMB are to be interpreted and supported makes a cear differentiation between impementation of video and audio services: The WordDMB Forum recommends that: DAB (ETSI EN ) or DAB+ (ETSI TS ) shoud be used for radio-centric services; DMB (ETSI TS ) shoud be used for services which incude a video component. Frank Herrmann s note: For video services based on DAB, DAB-IP might be empoyed as an aternative to DMB. EBU First pubished : Juy EBU TECNICAL REVIEW 25

26 CONTENT PROTECTION HDMI HDCP & the manufacturers perspective Dietrich Westerkamp Thomson, EICTA HDTV Issue Manager HDTV signas offer great opportunities to broadcasters, but there is aso the negative side a high risk of piracy. In order to protect prime content against iegitimate use, contentprotection mechanisms can be used. For the digita HDMI interface between an HDTV set-top box and an HD ready dispay device, HDCP technoogy is chosen. This is a too that can be used at the discretion of the broadcaster who can activate it by means of a switching signa. In the case of a piracy attack, the technoogy offers a revocation mechanism whereby a ist of revoked devices is transmitted in a safe way to the receiver, where it is stored. The avaiabiity of a content protection mechanism being a mandatory requirement of the EICTA HD ready ogo does not mean that the dispay device aways needs to be fed in a protected manner. Free-to-air signas that are transmitted in the cear are aways dispayed. The high quaity of digitay transmitted HDTV offers the broadcaster big opportunities but aso brings aong some risks not to be negected: pirates use the high-quaity signas to iegay copy them and start their own business, thereby negecting the copyright of the originator. One of the inks that are open to attacks is the digita baseband interface between a receiving set-top box (STB) and an HDTV dispay device. Here, either the Digita Visua Interface (DVI) [1] or the High- Definition Mutimedia Interface (HDMI) [2] is in use. In order to protect highquaity digita signas on these interfaces, a technoogy caed High-bandwidth Digita Content Protection (HDCP) [3] is used. The European CE industry association, EICTA [4], made HDCP part of their minimum requirements for an HDcapabe dispay device that is abeed with the HD ready ogo. This artice expains the function of HDCP and the way it is impemented. It aso highights the different positions of European broadcasters concerning the contro of the copy protection mechanism. As of today, the appication of any content-protection mechanism is mainy controed by the content owner. The broadcaster or pay-tv operator is obiged by its icence contracts to ensure adequate content protection by switching on an appropriate mechanism, and the receiving/recording/dispaying devices must have impemented it. High-bandwidth Digita Content Protection (HDCP) Fig. 1 sketches a digita transmission system for HDTV signas. The HDTV 26 EBU TECNICAL REVIEW 2007

27 CONTENT PROTECTION owners. The broadcasters as we as the equipment manufacturers are obiged to transmit the ists and react accordingy, based on the icence contracts they have signed for using HDCP. In order to protect these ists from being tampered with on their way to the receiver, they are transmitted with a digita signature. Figure 1 Concept diagram of a digita transmission system with Conditiona Access and HDCP copy protection of the dipay interface HDCP switchabe, programme-byprogramme signa from the head-end is sent to a set-top box (STB). In many cases a Conditiona Access (CA) system is used to enabe the protection of the content as we as the subscription management. Once the STB has received and decoded the signa, it needs to be forwarded to a suitabe dispay. In the case of HDTV signas, the digita connection between the STB and the dispay wi be either HDMI or DVI (Figs 2 and 3), with the former being the most up to date. If the content owner requests the broadcaster to protect the content against piracy, there must be a mechanism in pace that prevents someone from tapping the interface between the STB and the dispay and making an iega copy. For this purpose, the HDCP scheme has been deveoped. Using this mechanism, the content on the interface between the STB and the dispay device is scrambed in order to make it useess for pirates. Once a dispay device is hooked up to a source device (here, the STB), an initia authentication/negotiation procedure between the source and the dispay is started. In the course of this authentication procedure, keys are exchanged and vaidated, and the scrambing mechanism is activated. Authentication is aso needed in order to have the possibiity of taking action in case any of the devices invoved have been compromised in a way that coud be used for piracy. In those cases, the content owners can signa via so-caed revocation ists that the compromised devices are back-isted and sha no onger be permitted to transport signas using the HDCP scrambing mechanism. By this method, content owners can render such devices useess and hence pug the piracy hoes. The responsibiity for putting together these revocation ists is with the content Content protection on the dispay interface may not be needed for a the programmes broadcast by a particuar TV channe; there may even be TV channes that do not request any content protection. In those cases, the HDCP mechanism can be switched off and the content can be transmitted in the cear as a high-bitrate baseband video and audio signa. At present, such a switching mechanism is reaised within the different CA systems. In the same channe that transmits the programme in protected form, the information is transmitted to the STB whether any copy protection is needed on the dispay interface (DVI/HDMI). There are currenty various impementations in use that differ in their ways of controing the HDCP on/off switch. It goes without saying that contro over this switch is sensitive and wi not be made avaiabe to a potentia users of the STB incuding a potentia pirate! Figure 2 High-Definition Mutimedia Interface (HDMI): pug, socket and ogo (courtesy of HDMI.org) Figure 3 Digita Visua Interface (DVI): pug, socket and ogo (courtesy of DDWG.org) The way it is used is defined by the operator who specified the set-top box. In fact, the impementation in most cases is part of the Conditiona Access system impementation. Based on conditions set by the content owners, copy-contro mechanisms are even wider than the simpe on/off switching of HDCP on the digita interface. Amost a set-top boxes have anaogue as we as digita outputs, incuding one or more SCART pugs for hooking up standard-definition devices. In the case where HDCP on the digita interface is enabed (for protecting a 2007 EBU TECNICAL REVIEW 27

28 CONTENT PROTECTION Abbreviations A/D CA CE D/A DVB DVI EICTA Anaogue-to-Digita Conditiona Access Consumer Eectronics Digita-to-Anaogue Digita Video Broadcasting Digita Visua Interface European Information, Communications and HDCP HDMI SDTV STB Consumer Eectronics Technoogy Industry Association High-bandwidth Digita Content Protection High-Definition Mutimedia Interface Standard-Definition Teevision Set-Top Box proof device. One of the questions that needed to be answered was the necessity of impementing copy protection. The European CE, IT and communications industry association, EICTA, defined the HD ready and HD TV ogos (Fig. 4). Whie HD ready defines the minimum requirements for dispay devices, the HD TV ogo does the same for HDTV receiving equipment. Detais can be found on the EICTA website [4]. high-quaity HDTV signa), the anaogue interfaces may behave in severa different ways: They coud be copy-protected by an anaogue system but, at present, such systems ony exist for standard definition; The HD component interface coud be switched off, with ony the SD interface (SCART) deivering a copyprotected SDTV signa; A anaogue interfaces coud deiver a signa but ony in standard definition sometimes this can even be recordabe; A anaogue interfaces coud be switched off. It is very important to note that the behaviour of the anaogue interfaces is defined by the body that specifies the set-top box and has nothing to do with the HDCP mechanism described above. HDCP does not dea with any anaogue signas. Free-to-air content and copy protection Amost a HDTV set-top boxes on the European market are put there by pay-tv operators. At the end of 2006, there were approximatey STBs in consumer househods.this number is quicky heading towards one miion boxes, as further HDTV services get aunched in various European countries. An intense debate has occurred around the way these boxes shoud hande free-to-air content. A DVB set-top boxes defined for pay- TV are aso capabe of receiving freeto-air content. In the case of HDTV, the decoded signa is fed to the dispay device preferaby by the HDMI interface in order to best preserve the high quaity of the pictures. But the free-to-air broadcasters currenty have no infuence to contro the way HDCP is used (or not) on that interface. These rights are defined by the party that specified the set-top box the pay-tv operator. That being said, there is aso no obigation on free-to-air broadcasters to dea with the transmission of revocation ists. In fact, the existing boxes in Germany, the UK and France hande the HDCP switching differenty: some boxes eave HDCP on at a times whereas others switch HDCP on ony for specific programmes such as firstrun movies. In both cases, the free-to-air signas wi be dispayed on the connected HD-ready device and the viewer woud not even know whether copy protection is active or not. Obviousy there is one exception... once the dispay device has been misused for piracy activities and is consequenty put on the revocation ist, it wi not receive any further images when HDCP is switched on. HD ready and HD TV ogos and copy protection When HD-capabe dispay devices became avaiabe on the market pace, discussions started on which features needed to be impemented in order to have a future- Figure 4 EICTA ogos: (eft) HD ready for dispay devices and (right) HD TV for receiving device The HD ready minimum requirements incude anaogue as we as digita interfaces. The atter which coud be DVI or HDMI necessariy needs to have HDCP impemented. This was made mandatory in order to ensure that the consumer wi aways see an HDTV picture, even if the broadcaster or content provider decides to use copy protection on the output of the receiving device. After a engthy debate, EICTA decided not to make HDCP mandatory for a receiving equipment. This pays tribute to the fact that, in future, there might be free-to-air receivers without any CA system that simpy do not offer the technica means needed for HDCP impementation (i.e. a secure transmission channe for switching information and revocation ists). When the first HD ready devices came on the market, there was a campaign in the technica press that the HD ready ogo woud simpy be an industry action to have copy protection made mandatory in a cases. This definitey is not the case, because a interfaces aways accept signas that are offered without copy protection. However, the ogo assures the consumer that he wi aways see a picture uness 28 EBU TECNICAL REVIEW 2007

29 CONTENT PROTECTION Dietrich Westerkamp graduated from the University of Hannover. He then worked for severa years as a research assistant on data compression for images. Since 1985 he has been with Thomson, currenty as Director Standards Coordination. Over the past 20 years, Mr Westerkamp has been invoved in many projects to improve TV systems, ranging from HD-MAC and PALpus via DVC (the standard for digita camcorders) to MPEG and DVB. In the DVB Consortium, he has served for severa years as an industry member of the Steering Board. Within EICTA, he chairs the HDTV issue group that deveoped the HD ready and HD TV product ogos. his dispay device has been misused for piracy action and has been revoked. Handing of revocation ists In the current impementations, the revocation ist is stored in the STB. The receiving device gets the information via the broadcast channe as defined by the icensing authority, DCP LLC. Whenever a new version of the revocation ist is issued, the information stored in the receivers wi be updated. Concusions The HDMI interface is the best choice for deivering HDTV content from a receiving device to a modern dispay device. It maintains the quaity of the image at the highest possibe eve, by avoiding unnecessary cascaded A/D and D/A conversions. The high quaity of the signa on the interface makes it a target for signa pirates to make iega copies. HDCP is the means to prevent this. EICTA has made HDCP part of the minimum requirements for HD ready dispay devices in order to assure the consumer that he wi aways get a highquaity HDTV picture on his dispay. It needs to be underined here that the HDMI interface of the dispay aso accepts non-copy-protected signas. Once the connected set-top box uses HDCP a the time, free-to-air content wi aso aways be dispayed, even if it is transmitted via the copy-protected ink because the pay-tv operator who sponsored the set-top box has decided so. There is an ongoing debate at the eve of European standardization on whether there is a possibiity of defining a secure switching mechanism that woud aow every broadcaster to decide whether or not to activate HDCP. Looking at the current HD TV set-top boxes in the market pace, it can be seen ceary that they a impement HDCP and are using different concepts on how to contro the use of HDCP. Independent of that, a of these boxes can hande free-to-air signas and deiver them to the connected dispay. In a cases the consumer can enjoy the HDTV pictures uness he has misused his dispay device for piracy actions and the device has been put on the revocation ist. In that case, the screen wi remain dark. References [1] DVI DDWG, DVI Visua Interface, rev. 1.0, Apri 2, 1999 as further quaified in EIA/CEA-861 rev. B, A DTV Profie for Uncompressed High Speed Digita Interfaces May [2] HDMI HDMI Licensing, LLC, High- Definition Mutimedia Interface, rev. 1.3, November 10, [3] HDCP Inte, High-Bandwidth Digita Content Protection System, rev. 1.2, June 13, [4] EICTA Note from the Editor This artice outines the views of EICTA the European CE equipment manufacturers association on HD content protection using HDCP. The views of severa European broadcasters are presented in a separate artice pubished in this edition. EBU First pubished : October EBU TECNICAL REVIEW 29

30 CONTENT PROTECTION HDCP the FTA broadcasters perspective Jean-Pierre Evain European Broadcasting Union The first HD services have now been depoyed on pay-tv patforms using content-protection measures such as HDCP, in accordance with contractua obigations mandated by the production studios. Before ong, free-to-air TV patforms wi aso become invoved in HDCP. This artice provides technica information on the HDCP system, which is used to protect the HDMI ink from a set-top box to a dispay device (HDMI is the HDTV equivaent of the famiiar SCART connector used with standard-definition teevision). The artice aso expains what HDCP is and what it is not, and outines the views of severa different European broadcasters on methods for controing content protection. HDCP over HDMI: a de facto standard HDMI which has now supersed DVI in consumer eectronic products is a high-bandwidth interface between an HDTV transmitter (e.g. a set-top-box) and an HDTV repeater/receiver (e.g. a dispay device). Such interfaces are often referred to as dispay inks, with DVI more commony being found on persona computers. The HDMI interface can transmit HD digita video at bitrates up to 2.23 Gbit/s 1 at 720p or 1080i resoution, and up to eight channes of digita audio, 1. The current version of HDMI has a maximum bitrate imit of 4.95 Gbit/s but that figure wi be extended to 10.2 Gbit/s in a ater version of the interface. samped at 192 khz with 24 bits per sampe. Athough technicay chaenging, HDMI is ceary of interest to pirates for accessing high-quaity content sources in order to produce unauthorised copies. This is where HDCP comes in: it protects the content by encrypting the signa that is being carried over the HDMI (or, indeed, DVI) ink to the dispay device. HDCP is a proprietary technoogy from Inte Corporation, described in a specification that can be impemented under icence from the Digita Content Protection LLC (a subsidiary of Inte). The specification and icensing conditions can be found at As shown in Fig. 1, up to 128 devices can be used simutaneousy, provided that each piece of equipment is (1) HDCP-compiant and (2) recognized (authenticated) as a vaid secure impementation. In the context of broadcasting, the Upstream Content Contro Function is the signaing information deivered from the broadcast stream (e.g. DVB s free-to-air signaing for content protection and copy management CPCM). HDCP is based on inear Authentication and Key Exchange (AKE), a process famiiar to cryptoogists. The AKE process invoves the exchange of secret keys that are unique to each and every device. The authentication process assesses the vaidity of these keys incuding a revocation contro. If the AKE process succeeds, content is encrypted by the transmitter over the ink and deivered to the receiver which decrypts it according to rues securey set up during the authentication 30 EBU TECNICAL REVIEW 2007

31 CONTENT PROTECTION By whom? Figure 23 Buiding the FEC Frame process, and dispayed. If the AKE process fais, the dispay wi probaby remain back. Other options are possibe such as downscaing the content resoution, which doesn t seem to be widey impemented today. HDCP is a de facto standard as most manufacturers have icensed the technoogy from the Digita Content Protection LLC group and abide by contract to a certain number or impementation rues and obigations. DVB has adopted HDMI with HDCP as the associated protection mechanism. Furthermore, HDCP is mandated by EICTA in order to obtain the right to use the HD Ready ogo. HDCP content protection Why? The main reason for using HDCP is to prevent content being exposed and accessed in the cear, over high-bandwidth high-quaity digita interfaces from which materia coud be extracted e.g. to produce unauthorised copies. What? HDCP is a security too for content protection. It is not a copy management mechanism, used to carry and enforce usage restrictions. A copy management mechanism may in turn require the use of security toos such as HDCP to protect content. The fact that HDCP is activated has no other meaning than this content can ony be accessed by compiant and authenticated devices and sha not be subject to interpretation of derived usage restrictions (e.g. copy never or do not redistribute over the Internet ). It is essentia to understand, without any ambiguity, the precise nature and specific roe of HDCP. Exampe: Let s imagine an interface (e.g. other than HDCP) connecting a set-top-box to a PVR. In the case where copy never appies to some content, a compiant PVR wi not aow copying of this content, by means of deactivating the recording function. Conversey, content may be encrypted over the ink between the two devices to prevent tampering with it for unauthorised copying purposes. However, athough content might be protected over this ink, e.g. if no copy restriction appies, it sha sti be possibe to make a copy of this content. Hence content protection is not the same as copy management. The actua usage restriction associated with the activation of HDCP is unauthenticated access to content through this interface is not aowed. However, a content protection axiom woud state that HDCP shoud be activated whenever content is subject to a usage restriction. The decision to appy or not any content protection and copy management is the decision of the content owner, which subsequenty becomes a contractua obigation when content is icensed to service providers e.g. free-to-air or pay-tv broadcasters. Broadcasters are themseves often owners of the content that they produce and to which they may decide not to systematicay, if at a, appy content protection and copy management. One shoud know the potentia impications of activation or deactivation of HDCP on user access to protected content. The conditions under which HDCP might be used and how it might be used is subject to different circumstances and needs. As a first exampe, this artice focuses on free-to-air broadcasting but it is interesting to note that certain pay-tv operators wish to have the fexibiity to activate HDCP on a content-by-content basis, whie it is deactivated by defaut! Other pay-tv operators have specified their proprietary set-top-box boxes with HDCP being activated by defaut. As far as free-to-air is concerned, different positions have been expressed that correspond to different market and reguatory situations: Scenario 1 Free-to-air (FTA) or cear-to-air (CTA). In both cases, access is granted but imited to a particuar geographica ocation when FTA content is deivered in scrambed form. FTA content that has been protected for deivery can remain protected after acquisition through the activation of HDCP, which coud occur through signaing in the conditiona access system (as for pay- TV), or by defaut in the receiver. There is aso a need to be abe to deactivate HDCP (and subsequenty any simiar content protection mechanism) for some content. Content coud remain in the cear after geographica deivery uness otherwise instructed through proper DVB free-to-air signaing information EBU TECNICAL REVIEW 31

32 CONTENT PROTECTION Abbreviations AKE CPCM CTA DCP-LLC DTCP DVB Scenario 2 Authentication and Key Exchange (DVB) Content Protection and Copy Management Cear-To-Air Digita Content Protection LLC icensing group Digita Transmission Copy Protection Digita Video Broadcasting DVI EICTA FTA HDCP PVR SRM When? Digita Visua Interface European Information, Communications and Consumer Eectronics Technoogy Industry Association Free-To-Air High-bandwidth Digita Content Protection Persona Video Recorder System Renewabiity Message If the do_not_scrambe fag is set to true, HDCP shoud be deactivated. It is acknowedged that, athough originay designed to contro DVB Content Protection and Copy Management (DVB CPCM) scrambing, this signaing shoud equay appy to HDCP and simiar protection mechanisms independenty of the impementation of DVB CPCM. But when does it reay become essentia to contro content protection over a highbandwidth dispay ink? The answer to that question ies principay in two key impementation features of HDCP, i.e. egacy compiance and revocation. For CTA content deivered in the cear, some EBU members want HDCP being deactivated by defaut on CTA-capabe devices. If a set-top-box gives access to CTA content and pay-tv content, independenty of each other, it shoud be possibe to activate or deactivate HDCP according to the defaut state originay set uness otherwise instructed through proper DVB free-to-air signaing information. HDCP deactivation shoud preferaby be the defaut condition for such CTA set-top boxes in a horizonta market. Scenario 3 Some CTA broadcasters woud prefer HDCP being activated by defaut with the fexibiity to deactivate it for certain content through proper DVB free-to-air signaing information. Scenario 4 If FTA/CTA content is deivered as part of a pay-tv service to pay-tv set-top-boxes, the defaut HDCP state wi be defined by the pay-tv operator as we as the possibiity and mechanisms to activate or deactivate HDCP. The above vaid, but diverse, scenarios iustrate the need for HDCP (and simiar content protection mechanisms) to be switchabe on a content-bycontent basis from one initia state (either on or off by defaut) to another. It seems ogica to activate HDCP content protection when usage restrictions such as imited access, copying, redistribution and consumption appy, because unauthenticated access to content in the cear woud aow circumventing these restrictions. Conditiona Access (CA) systems can pay the roe of Upstream Content Contro Function that activates or deactivates HDCP content protection. In some cases, the simpe fact that content is deivered in a scrambed form is sufficient to require the activation of HDCP. In other CA configurations, the same channe aso carries usage restriction messages, which aows more fexibiity such as the activation of HDCP on a content-bycontent basis in set-top-boxes with HDCP off by defaut, or for deactivating HDCP for FTA content after acquisition. DVB considers that CTA content sha be considered as protected as ong as DVB free-to-air signaing is deivered aongside this content within the broadcast stream. DVB has specified free-to-air signaing to aow or prevent: 1) the redistribution of content over the Internet (contro_remote_access_ over_the_internet); 2) the scrambing of content (do_not_ scrambe); 3) the use of revocation ists (do_not_ appy_revocation). HDCP compiance In a perfect word where a devices are HDCP compiant, the norma honest user experience woud be unaffected by content fowing over the HDCP interface in a scrambed form or not. But there wi be a egacy of eary adopters with dispays without HDCP or, not to be underestimated, dispays with eary and not fuy-compiant HDCP impementations. One of the reasons pay-tv operators switch HDCP off by defaut may have been to ensure access to owners of eary dispays and to overcome potentia eary interoperabiity probems. FTA broadcasters shoud share the same concern. The evoution of the HDCP specification might generate a new egacy... and, in particuar, a greater interoperabiity chaenge managing the revocation ists. The revocation diemma In a fuy HDCP-compiant word, having protection on by defaut woudn t be such an issue if there weren t the additiona burden of revocation which, in turn, woud be ess probematic if 32 EBU TECNICAL REVIEW 2007

33 CONTENT PROTECTION managed on a content-by-content basis as recommended by DVB. But HDCP (and other simiar protection mechanisms such as DTCP) currenty makes this more compicated. Revocation consists of identifying devices that have been compromised and coud be misused as a sink to access content and generate unauthorised copies. A device is compromised when (1) a device private key has been coned and repicated in pirate devices or (2) the private key of that device has been made pubic (e.g. after being ost or stoen). Compromised devices are identified by their individua keys, compied into revocation ists which are typicay distributed with the content (in the signa or with removabe medias) in signed / authenticated System Renewabiity Messages (SRMs) but can aso be embedded into new devices. This ist is consuted during the HDCP authentication procedure and athough the AKE process is successfu, a device woud not be granted access to content if backisted. The Content Participant Agreement defines the conditions under which content owners who have signed the agreement may request revocation of devices. The responsibiity for putting together these revocation ists is with the content owners. Broadcasters are obiged to transmit the ists and react accordingy by the icence contracts they have signed for HDCP. Athough version 1.1 of the HDCP specification was not specific about revocation ist management, version 1.2 defines a device-based revocation mechanism. This means that revocation ists must be permanenty stored into devices. Revocation ists are updated each time a device receives a more recent ist either with the content or when interconnected with another device (e.g. a new device with a preoaded revocation ist) either directy or through a home network. According to this specification, revocation is per device and not per content. SRMs are signed using a pubic key deivered by the Digita Content Protection LLC group. They do not require particuar protection to be transmitted. FTA/CTA broadcasters shoud be asked to coaborate in the deivery of such ists if they require the activation of HDCP. A buffer of 5 KBytes restricts the number of keys that can be stored in a device to one Vector Revocation List (the individua 40-bit keys of 128 devices), which has a imiting effect on the bandwidth needed to carry the SRMs and its cost for broadcasters. One key of one device can actuay deactivate thousands of devices sharing a compromised key. Crypto-anaysis has demonstrated that HDCP coud be considered broken if 40 keys are compromised. A new version is in preparation, which woud justify the handing of more than 128 devices, as envisaged in the HDCP specification. But the use of this new version may raise compatibiity and egacy issues. Why is device revocation dangerous for FTA broadcasters? If a receiving device that gives access to both free-to-air and pay-tv services has been instructed to backist some equipment (e.g. a dispay) for pay-tv content, then per device revocation woud resut in turning the screen back for pay-tv but aso free-to-air services. In this context, the back-screen threat is not in favour of HDCP being set on by defaut. However, a soution has been agreed within DVB by defining the free-to-air signaing fag do_not_appy_ revocation, which aows deactivating revocation on a per content basis for the associated FTA/CTA content. Obviousy, this soution requires being impemented by HDCP to be effective. Summary Like pay-tv operators, FTA/CTA broadcasters across Europe see different possibe uses of HDCP but woud ike the fexibiity to activate or deactivate it on a per content basis. This is a requirement aready endorsed by DVB for more generic content protection and copy management. HDCP is ony content protection and not a copy management scheme. Usage restrictions cannot be derived or interpreted from the activation of HDCP but, in principe, HDCP woud be activated when usage restrictions appy to content. Jean-Pierre Evain joined the EBU s Technica Department in 1992 to work on New Systems and Services, having spent six years in the R&D aboratories of France-Téécom (CCETT) and Deutsche Teekom. Mr Evain manages a EBU metadata activities. He represents the EBU in severa DVB groups regarding metadata as we as Copy Protection and Digita Rights Management. He aso represents the EBU in the IPTC consortium (news metadata) EBU TECNICAL REVIEW 33

34 CONTENT PROTECTION HDCP is a de facto standard that has been impemented differenty in various proprietary impementations for pay-tv. Meeting the needs of FTA broadcasters in the ong term, in a horizonta market, may require some adaptation to those currenty deveoped for pay-tv. In a fuy HDCP-compiant word, having content protection on by defaut woud not be a probem, notwithstanding the additiona burden of revocation. This in turn woud be ess probematic if managed on a per content basis. But HDCP (and other simiar protection mechanisms such as DTCP) has opted for device-based revocation. In such conditions, pay-tv set-top-boxes that are revoked to protect pay-tv premium content wi no onger deiver FTA content to users uness using the DVB FTA switching fag. This must not prevent FTA broadcasters being invoved in the revocation decision-making process to counter-baance the market impact of such actions. FTA broadcasters woud be asked to coaborate in the deivery of revocation messages if they require the activation of HDCP. DVB has agreed a free-to-air signaing scheme, which offers a soution to severa of the key issues mentioned in this artice and, more particuary, concerning HDCP activation and per content revocation. It is strongy advised that future HDCP impementations respond to such signaing, if not aready. One issue of serious concern to potentia FTA broadcaster-users of HDCP is the ack of stabiity of the specification. The specification has aready changed from version 1.1 to version 1.2 and 1.3. There are critica egacy and interoperabiity issues. The vaue of HDCP wi be weakened if the specification and compiance rues are being changed without open consutation. References 1. High-Bandwidth Digita Content Protection System, revision 1.1, 9 June High-Bandwidth Digita Content Protection System, revision 1.2, 13 June High-Bandwidth Digita Content Protection System, revision 1.3, 21 December Conditions for High Definition Labeing of Dispay Devices, 19 January 2005 ( Note from the Editor This artice outines the views of EICTA the European CE equipment manufacturers association on HD content protection using HDCP. The views of severa European broadcasters are presented in a separate artice pubished in this edition. EBU First pubished : October EBU TECNICAL REVIEW 2007

35 VIDEO STREAMING Mutipe Description Coding Andrea Vitai STMicroeectronics a new technoogy for video streaming over the Internet The Internet is growing quicky as a network of heterogeneous communication networks. The number of users is rapidy expanding and bandwidth-hungry services, such as video streaming, are becoming more and more popuar by the day. However, heterogeneity and congestion cause three main probems: unpredictabe throughput, osses and deays. The chaenge is therefore to provide: (i) quaity, even at ow bitrates, (ii) reiabiity, independent of oss patterns and (iii) interactivity (ow perceived atency)... to many users simutaneousy. In this artice, we wi discuss various we-known technoogies for streaming video over the Internet. We wi ook at how these technoogies partiay sove the aforementioned probems. Then, we wi present and expain Mutipe Description Coding which offers a very good soution and how it has been impemented and tested at STMicroeectronics. Packet networks [1] [2] Heterogeneity adds up with errors and congestion: backbone and wired inks have an increasing capacity whie, at the same time, more and more ow-bandwidth error-prone wireess devices are being connected. Throughput may become unpredictabe. If the transmission rate does not match the capacity of the botteneck ink, some packets must be dropped. The deivery system may provide prioritisation: the most important packets are given a preferentia treatment, whie the east important packets are dropped first. However, usuay networks wi drop packets at random. Packet oss probabiity is not constant; on the contrary, it can be widy varying, going from very good (no oss) to very bad (transmission outages). This makes the design of the deivery system very difficut. Usuay there are two options: the system can be designed for the worst case; or it can be made adaptive. If it is designed for the worst case, it wi be inefficient every time the channe is better than the worst case, i.e. most of time. Conversey, if it is designed to be adaptive, it wi most probaby adapt too ate. Data-independent content deivery technoogies ARQ: Automatic Repeat request One of the most effective techniques for improving reiabiity is the retransmission of ost packets: Automatic Repeat request, or ARQ. TCP-based content deivery is based on this EBU TECNICAL REVIEW 35

36 VIDEO STREAMING If osses are sporadic, this technique is very efficient: packets are successfuy sent ony once. On the other hand, if osses are frequent, retransmissions can even increase congestion and aso the oss rate, a vicious cyce (this is avoided in TCP-based content deivery). Retransmission is very usefu in pointto-point communications where a feedback channe is avaiabe. However, when broadcasting to many receivers, the broadcaster cannot hande a the independent retransmission requests. The added deay of the retransmission is at east one round-trip transport time. But each retransmission can aso be ost, and the deay can be arbitrariy arge. This is critica for streaming video: the deay of a retransmitted packet may be much onger than inter-arriva times and, as a consequence, streaming may suffer stas. This deay adds up in the receiver buffer which must be arge enough to compensate for variation in the interarriva times (jitter). FEC: Forward Error Correction / Erasure Recovery Another very effective technique is channe coding, i.e. the transmission of redundant packets that aow recovery of erroneous / ost packets at the receiver side: Forward Error Correction / Erasure Recovery, or FEC. If the oss rate is known, the added redundancy can be made just enough to compensate. Unfortunatey, in the rea word not ony the amount of osses is not known, but aso it is widy timevarying. This, couped with the fact that this technique has an a-or-nothing performance, makes its use very difficut: either errors are too much or they are ess than expected. If osses are too much, the recovery capabiity wi be exceeded. Added redundancy wi not be enough and the osses wi not be recovered. Decoded quaity wi be very bad (ciff effect). Because of this, to be safe, broadcasters typicay consider the worst case and choose to increase the amount of redundancy at the expense of the video. The video is compressed more heaviy, owering the fina decoded quaity. If errors are ess than expected, which is probabe when the system is designed for the worst case, the osses wi be recovered. The decoded quaity wi be guaranteed, unaffected by oss patterns. However capacity is wasted: ess redundancy coud be used eaving room for a higher-quaity ighty-compressed video. Adaptation coud be used in principe to dynamicay baance the added redundancy and video compression, but it is rarey done because of the difficuty. Decoded quaity is ower than it is possibe to get. The compexity can be very high: encoding and decoding of redundant packets requires memory and computationa power. Efficient schemes for error correction / erasure recovery require processing of a arge number of video packets. Therefore the added deay is not arbitrariy arge, but it can be significant. Data-dependent content deivery technoogies Robust source coding The more efficient the video encoder, the more important a video packet is. When compression efficiency is very high, the oss of a packet has potentiay a devastating effect. Then, a heavy recovery mechanism, such as compex FEC codes, must be used to reduce the probabiity of this happening. Conversey, when the compression efficiency is ow, the oss of a packet has itte effect. In this case, conceament techniques do exist that can reduce or even competey hide the effect of the oss. In this case, a ight recovery mechanism can be used. Therefore, compression efficiency shoud be tuned carefuy, taking into account the effect of osses, the effectiveness of conceament techniques and the effectiveness of the recovery mechanism. The avaiabe bandwidth can then be optimay spit between the video data and redundant data. Said in other words, it is aways usefu to optimize the parameters of the source encoder and of the channe encoder jointy (a technique known as joint source-channe coding ). In the case of mutimedia communications, this means expoiting the error resiience that may be embedded in compressed mutimedia bitstreams, rather than using compex FEC codes or compex communication protocos. Video encoders use a bunch of techniques to efficienty squeeze the video: prediction (aso known as motion estimation and compensation), transform, quantization and entropy coding. Prediction is one of the most important techniques from the point of view of compression efficiency: the current video is predicted from the previousy transmitted video. Because of this, video packets are dependent on previous packets. If these packets have not been successfuy received, then the current packet is not usefu. This is known as oss propagation. Compression efficiency can be a trade-off for robustness by reducing the amount of prediction (i.e. more intra coding): dependencies wi be reduced, stopping the oss propagation effectivey. Transmission deay can aso be a tradeoff for robustness. Video packets can be reorganized (in so-caed intereaving buffers ) so that consecutive video packets do not represent neighbouring video data. This is done to deocaise the effect of osses and ease the conceament. A ong burst of ost packets wi affect portions of the video which are far apart from each other. Lost portions can then be conceaed effectivey by expoiting neighbouring video data. Conceament is usuay done bindy at the receiver side. However, the transmitter 36 EBU TECNICAL REVIEW 2007

37 VIDEO STREAMING can encode hints (conceament data) that increase its effectiveness. Obviousy this consumes part of the avaiabe bandwidth. A these techniques are very effective, but it is very difficut to choose an optima set of parameters. It is especiay difficut when there are many receivers which experience different channe conditions. Mutipe Description Coding [3] [4] Mutipe Description Coding (MDC) can be seen as another way of enhancing error resiience without using compex channe coding schemes. The goa of MDC is to create severa independent descriptions that can contribute to one or more characteristics of video: spatia or tempora resoution, signa-to-noise ratio, frequency content. Descriptions can have the same importance (baanced MDC schemes) or they can have different importance (unbaanced MDC schemes). The more descriptions received, the higher the quaity of decoded video. There is no threshod under which the quaity drops (ciff effect). This is known as gracefu degradation. The robustness comes from the fact that it is unikey that the same portion of the same picture is corrupted in a descriptions. The coding efficiency is reduced depending on the amount of redundancy eft among descriptions; however channe coding can indeed be reduced because of enhanced error resiience. Experiments have shown that Mutipe Description is very robust: the deivered quaity is acceptabe even at high oss rates. Descriptions can be dropped wherever it is needed: at the transmitter side if the bandwidth is ess than expected; at the receiver side if there is no need, or if it is not possibe to use a descriptions successfuy received. This is known as scaabiity. It shoud be noted that this is a side benefit of Mutipe Description Coding which is not designed to obtain scaabiity; instead it is designed for robustness. Descriptions of the same portion of video shoud be offset in time as much as possibe when streams are mutipexed. In this way a burst of osses at a given time does not cause the oss of the same portion of data in a descriptions at the same time. If intereaving is used, the same criterion is to be used: descriptions of the same portion of video shoud be spaced as much as possibe. In this way a burst of osses does not cause the oss of the same portion of data in a descriptions at the same time. The added deay due to the amount of offset in time, or the intereaving depth, must be taken into account. Layered Coding Layered Coding (LC) is anaogous to Mutipe Description Coding (MDC). The main difference ies in the dependency. The goa of LC is to create dependent ayers: there is one base ayer and severa enhancement ayers that can be used, one after another, to refine the decoded quaity of the base ayer. Layers can be dropped wherever required but they cannot be dropped at random: the ast enhancement ayer shoud be dropped first, whie the base ayer must never be dropped. If the base ayer is not received, nothing can be enhanced by the successive ayers. Layered Coding is designed to obtain this kind of scaabiity. Repair mechanisms are needed to guarantee the deivery of at east the base ayer. Moreover: because of the unequa importance of ayers, repair mechanisms shoud unequay protect the ayers to better expoit Layered Coding. However not a networks offer this kind of services (prioritization). Recovery mechanisms and Layered / Mutipe Description Coding Channe coding is needed by Layered Coding. However channe coding can aso be used with Mutipe Description Coding. Generay speaking, it is better to adapt the protection eve of a given description / ayer to its importance, a technique commony known as unequa error protection. Unequa error protection is better even in the case of equay-important descriptions (baanced MDC). In fact, armouring ony one description may be more effective than trying to protect a descriptions. If this is done, there is one description which is heaviy protected. If the channe becomes reay bad, this description is ikey to survive osses. Then the decoder wi be abe to guarantee a basic quaity, thanks to this description. Abbreviations ARQ Automatic Repeat request FEC Forward Error Correction IF-PDMD Independent Fux Poyphase Downsamping Mutipe Description LC Layered Coding MD Mutipe Description MDC Mutipe Description Coding TCP Transmission Contro Protoco Summary of reviewed technoogies and their characteristics To summarize, here is an overview of the technoogies that can be used for video streaming over packet networks, to compensate for osses due to errors and congestion: 2007 EBU TECNICAL REVIEW 37

38 VIDEO STREAMING Data-independent content deivery technoogies Automatic Repeat Request (ARQ): suitabe ony for point-to-point, needs feedback, added deay arbitrariy arge. Forward Error Correction (FEC): no feedback required, a-or-nothing performance (ciff effect), waste of capacity when tuned for worst case, compexity, significant added deay. Data-dependent content deivery technoogies Robust Source Coding: difficut to choose optima parameters Mutipe Description Coding (MDC): no ciff effect (gracefu degradation), no prioritisation needed, aows scaabiity, very robust even at high oss rates Layered Coding (LC): requires prioritisation and recovery mechanisms, aows efficient scaabiity It shoud be noted that packet networks are designed to deiver any kind of data: a data-independent technique is therefore aways needed. The best option is Forward Error Correction / erasure recovery (FEC). For mutimedia data, such as video (and audio as we), severa smart techniques exists. In this case the best option is Mutipe Description Coding (MDC). Standard-compatibe Mutipe Description Coding [6] [8] Losses due to errors and congestion do cause visibe artefacts in decoded video: oss conceament techniques may hep, but they are rarey effective, as can be seen in Fig. 1. This expains the need for an effective technique to recover osses and/or ease the conceament. Figure 1 On the eft, errors are not conceaed. On the right, state-of-the-art conceament has been appied Automatic Repeat request (ARQ) is suitabe ony for point-to-point communications and cannot be easiy scaed to broadcast scenarios; furthermore, it requires a feedback channe which may not be avaiabe. FEC is effective ony if compex (which means: more power, deay, etc) and it has a threshod which yieds an a-ornothing performance (the ciff effect). Robust source coding is difficut to use, as parameters are difficut to be tuned. Layered Coding is not designed for robustness and reies on the aforementioned recovery mechanisms. Conversey, Mutipe Description Coding does not require a feedback channe and does not have an a-or-nothing behaviour: instead it has gracefu degradation (more descriptions, more quaity), pus it offers free scaabiity (to transmit as many descriptions as possibe, receive as many as needed). The question is: if Mutipe Description Coding does serve the purpose we (robustness, effectiveness), then what is the price to be paid when impementing this soution (efficiency, bandwidth, quaity, compexity, compatibiity with egacy systems). Standard compatibiity It is not easy to design and impement a Mutipe Description video coding scheme. There are many estabished video coding standards depoyed in the rea word: e.g. MPEG-2, MPEG-4, H.263 and H.264. It is difficut to impose yet another standard which is more compex. There are many other techniques avaiabe for creating mutipe descriptions: mutipe description scaar or vector quantization, correating transforms and fiters, frames or redundant bases, forward error correction couped with ayered coding, spatia or tempora poyphase downsamping (PDMD). The best choice can be found by foowing this criteria: Compatibiity: the possibiity to use standard encoders for each description and the possibiity of being compatibe with egacy systems; Simpicity: minimum added memory and computationa power; Efficiency: for a given bandwidth and when there are no osses, the minimum oss of decoded quaity with respect to the best quaity deivered by standard coding. Among the aforementioned techniques, poyphase downsamping is particuary interesting as it is very simpe and it can be easiy impemented using standard state-of-the-art video encoders. The sequence to be coded is subdivided into mutipe subsequences which can then be coded independenty. This is done in a pre-processing stage 38 EBU TECNICAL REVIEW 2007

39 VIDEO STREAMING (Fig. 2). At the decoder side, there is a post-processor stage (Fig. 3) in which decoded subsequences are merged to recreate the origina one. This simpe scheme is aso known as Independent Fux Poyphase Downsamping Mutipe Description coding (IF-PDMD). This scheme is competey independent of the underying video encoder. Subdivision to create descriptions can be done aong the tempora axis (e.g. by separating odd and even frames) or in the spatia domain (e.g. by separating odd and even ines). As encoding of each description is independent from others, there can be sight differences in the decoded quaity. When tempora subdivision is used a potentiay annoying artefact may arise: the difference among odd and even frames may be perceived as fashing. Figure 2 Pre-processing stage: downsamping in spatia domain. Odd and even ines are separated, the same is done for coumns. Four descriptions are created. Figure 3 The whoe chain: pre-processing, encoding, transmission, decoding, post-processing On the contrary, when spatia subdivision is used (see Fig. 4), a potentiay peasant artefact may arise: the difference between descriptions may be perceived as dithering, a known technique appied in graphics to hide encoding noise. Spatia subdivision has two more advantages: Two descriptions can be created by separating odd and even ines: interaced video is then reduced to two smaer progressive video streams which may be easier to encode. Four descriptions can be created by separating odd and even ines, and then separating odd and even coumns: high definition video (HDTV) is then reduced to four standard definition video streams which can be encoded using existing encoders. It shoud be noted that keeping Mutipe Description Coding decouped from the underying codec prevents it from giving its best. To get maximum quaity and to encode the descriptions with east effort, joint or coordinated encoding coud be used. Aso, to expoit the redundancy and to maximize the error resiience, joint Mutipe Description decoding is recommended. As an exampe, video encoders can share expensive encoding decisions (motion vectors) instead of computing them; aso they can coordinate encoding decisions 2007 EBU TECNICAL REVIEW 39

40 VIDEO STREAMING Moreover, MD encoding can even be beneficia. In fact, mutipexed descriptions can be marked so that od decoders beieve that they are mutipe copies of the same sequence. Figure 4 Dithering effect as a resut of spatia downsamping: 4 descriptions are created by separating odd/even ines and taking every other pixe. As encoding of each description is independent from others, the decoded quaity may differ sighty. As an exampe, when four descriptions are transmitted, the od decoder wi beieve that the same video packet is transmitted four times. Actuay, they are four sighty different packets, but this does not matter. The decoder can be instructed to decode ony the first copy and, if this copy is not received correcty, it can be instructed to decode another copy. Why use Mutipe Description Coding? Firsty: increased error resiience. Secondy: we get scaabiity for free. Robustness Mutipe Description Coding is very robust, even at high oss rates (see Fig. 6). It is unikey that the same portion of a given picture is corrupted in a the descriptions. It s as simpe as that! Figure 5 30% packet oss; eft: the output of a standard decoder, not aware of Mutipe Description, has been instructed to see descriptions as repicas of the same packet (fake standard encoding); right: the output of a Mutipe Description decoder. A more sophisticated point of view is to note that descriptions are intereaved. In fact, when the origina picture is reconstructed, descriptions are merged by intereaving pixes. A missing portion in one description, wi resuts in scattered missing pixes. These pixes can easiy be estimated by using neighbouring avaiabe pixes. (quantization poicies) to enhance the quaity or resiience (intereaved mutiframe prediction poicies, intra-refresh poicies). Decoders can share decoded data to ease error conceament; aso they can share critica interna variabes (anchor frame buffer) to stop error propagation due to prediction. It is worth mentioning that, if baanced descriptions are propery compressed and packed, any osses can be recovered before the decoding stage. In this case, decoders are preceded by a specia processor in which ost packets are recovered by copying simiar packets from other descriptions. Simiar packets are those that carry the same portion of video data. The scheme is aso compatibe with systems not aware of Mutipe Descriptions (see Fig. 5). In fact, each description can be decoded by a standard decoder, which need not be MD-aware in order to do this. Of course, if spatia MD has been used, the decoded frame has a smaer size... whie if tempora MD has been used, the decoded sequence has a ower frame rate. It is assumed that errors are independent among descriptions. This is true ony if descriptions are transmitted using mutipe and independent channes. If one singe channe is used instead, descriptions have to be suitaby mutipexed. If this is done, error bursts wi be broken by the demutipexer and wi ook random, especiay if the burst ength is shorter than the mutipexer period. Scaabiity There are many scenarios where scaabiity can be appreciated. With mobie terminas 40 EBU TECNICAL REVIEW 2007

41 VIDEO STREAMING Figure 6 Same aggregate bandwidth, number of packets and average packet size, and with 30% packet oss rate. Top row: standard coding. Bottom row: four mutipe descriptions generated by separating odd/even ines and taking every other pixe; before and after conceament. in mind, when standard coding is used, the whoe bitstream shoud be decoded and downsized to adapt it to the sma dispay. Power and memory are wasted. Conversey, when Mutipe Description is used, a termina can decode ony the number of descriptions that suits its power, memory or dispay capabiities. Aso, when the channe has varying bandwidth, it woud be easy to adapt the transmission to the avaiabe bandwidth. Descriptions may simpy be dropped. Instead, a non-scaabe bitstream woud require an expensive transcoding (reencoding the video to fit the reduced avaiabe bitrate). a higher priority or is more protected than the enhancement ayer. When MD coding is used, there is no base ayer. Each description can be decoded and used to get a basic quaity sequence. More decoded descriptions ead to higher quaity. There is no need to prioritise or protect a bitstream. Finay, it must be noticed that at very ow bitrates the quaity provided by Mutipe Description Coding is greater than that provided by standard coding. This happens because the ow bitrate target can easiy be reached by simpy dropping a descriptions except one. On the contrary, with standard coding a rough quantization step must be used. Artefacts introduced This kind of scaabiity shoud be compared to the scaabiity provided by Layered Coding: think about osing the base ayer whie receiving the enhancement. It happens that the received enhancement is useess and bandwidth has been wasted. Usuay, in order to avoid this, the base ayer is given Figure 7 Site foreman, CIF resoution (352x288 pixes) at 155 kbit/sec using MPEG-4/10 encoder. Standard coding on the eft, one out of four mutipe descriptions on the right EBU TECNICAL REVIEW 41

42 VIDEO STREAMING by heavy quantization are more annoying than artefacts introduced by dropping descriptions (see Fig. 7) Foeman, Poss=0.1 Burst=10 Why not use Mutipe Description Coding? At a given bitrate budget, there is a quaity oss with respect to standard (singe description) coding. The oss depends on the resoution (the ower the resoution, the higher the oss) and on the number of descriptions (the more the descriptions, the higher the oss). Descriptions are more difficut to encode. Prediction is ess efficient. If spatia downsamping is used, pixes are ess correated. If tempora downsamping is used, motion compensation is not accurate because of the increased tempora distance between frames. Aso, syntax is repicated among bitstreams. Think about four descriptions. There are four bitstreams. Each hods data for a picture which has 1/4th the origina size. When taken a together, the four bitstreams hod data for the same quantity of video data as the singe description bitstream. The bit-budget is the same. However, the syntax is repicated, therefore there is ess room for video data. However, it must be noted that it is not fair to compare the decoded quaity of Mutipe Description Coding with standard (singe description) coding when there are no osses. Standard coding has been designed for efficiency whie Mutipe Description Coding has been designed for robustness. If there are no osses, this increased error resiience is useess. A fair comparison woud be to compare error-resiient standard coding with Mutipe Description Coding. As an exampe, the standard bitstream can be made more error resiient by reducing the amount of prediction (increased intra refresh). The intra refresh shoud be increased unti the quaity of the decoded video is PSNR (db) Figure 8 Quaity frame-by-frame: the back ine corresponds to standard coding protected by Reed-Soomon forward error correction (a-or-nothing behaviour), the bue ine corresponds to two Mutipe Descriptions (sighty ower average quaity, but much ower variance). equa to the quaity of decoded Mutipe Description. Then it woud be possibe to evauate the advantage of using Mutipe Description by etting the packet oss rate increase and see which coding is better. Experiments have shown [5] that Mutipe Description is sti superior when compared to error-resiient standard coding, even if the packet oss rate is very ow (~1%). Simuations have been done at the same aggregate bitrate and same decoded quaity using one of the most efficient FEC schemes: Reed-Soomon (R-S) codes (see Fig. 8). From a higher point of view, we might decide to reduce channe coding and use part of its bit-budget for Mutipe Descriptions bitstreams, therefore increasing the quaity of the decoded Mutipe Descriptions. Foreseen appications of Mutipe Description Coding Divide-and-rue approach to HDTV distribution: HDTV sequences can be spit into SDTV descriptions; no custom high-bandwidth is required Frame number Easy picture-in-picture: with the cassica soution, a second fu decoding is needed pus downsizing; with MDC/LC, it is sufficient to decode one description or the base ayer and paste it on the dispay. Adaptation to ow resoution/ memory/power: mobies decode as many descriptions/ayers as they can based on their dispay size, avaiabe memory, processor speed and battery eve. pay-per-quaity services: the user can decide at which quaity eve to enjoy a service, from ow-cost ow-resoution (base ayer or one description ony) to higher cost high-resoution (by paying for enhancement ayers / more descriptions). Easy ce hand-over in wireess networks: different descriptions can be streamed from different base stations expoiting muti-paths on a ce boundary. Adaptation to varying bandwidth: the base station can simpy drop descriptions/ayers; more users can easiy be served, and no trans-coding process is needed. 42 EBU TECNICAL REVIEW 2007

43 VIDEO STREAMING Muti-standard support (simucast without simucast): descriptions can be encoded with different encoders (MPEG-2, H.263, H.264); there s no waste of capacity as descriptions carry different information. Enhanced carouse: instead of repeating the same data over and over again, different descriptions are transmitted one after another; the decoder can store and combine them to get a higher quaity. Appication to P2P (peerto-peer) networks In P2P networks users hep each other to downoad fies. Each fie is cut into pieces. The more popuar a fie is, the greater the number of users that can support a given user by transmitting the missing pieces. Streaming however is a different story. The media cannot easiy be cut into pieces, and in any case the pieces shoud be received in the correct order from a given user to be usefu for the payout. Aso, a typica user has greater downink capacity than upink capacity. Therefore (s) he is not abe to forward a the data (s) he receives and cannot hep other users that are wiing to receive the same stream. One of the most effective soutions for ive streaming has been impemented by Octoshape [7]. This is their scheme: A video that woud require 400 kbit/s is spit into four streams of 100 kbit/s each. Therefore N redundant 100 kbit/s streams are computed, based on the origina four streams; the user is abe to reconstruct the video given any four streams out of the avaiabe streams (the four origina and the N redundant streams) this can be done using an (N,4) Reed-Soomon FEC. Foowing this scheme, the typica user is abe to fuy use the upink capacity even if it is smaer than the downink capacity. Each user computes and forwards as many redundant streams as possibe, based on the capacity of its upink. A very simiar scheme can be impemented using Mutipe Description Coding: Four descriptions can be created by separating odd and even ines and taking every other pixe; each subsequence is encoded in 1/4 of the bitrate that woud have been dedicated to fu resoution video. Redundant descriptions can be created by further processing video data; as an exampe: averaging the four aforementioned descriptions, and so on. This is known as frame expansion. Frame expansion can easiy be expained by this simpe exampe: 2 descriptions can be generated by separating odd and even ines as usua; a 3rd description can be generated by averaging odd and even ines. It is cear that perfect reconstruction (except for quantization noise) is achieved if any 2 descriptions out of 3 are correcty received. Frame expansion can be seen as equivaent to a Forward Error Correction code with rate 2/3: one singe erasure can be fuy recovered (except for the quantization noise). However, unike FEC, there is no threshod: if there is more than one erasure, received descriptions are sti usefu. Moreover, the redundancy can be controed easiy by quantizing the third description more heaviy. Concusions Two data independent content deivery techniques have been presented: Automatic Repeat request (ARQ) and Forward Error Correction (FEC). The atter is preferabe as it does not require a feedback from receivers and is then suited to broadcast. However this technique has an a-or-nothing performance: when the correction capabiity is exceeded the quaity of decoded video drops. Three data dependent content deivery techniques have been presented: robust source coding, Mutipe Description Coding (MDC) and Layered Coding (LC). The atter is aso known as Scaabe Video Coding (SVC) as it aows efficient scaabiity: ayers can be decoded one after another, starting from the base ayers; ayer have different importance and require prioritisation Andrea L. Vitai graduated in Eectronics at the Mian Poytechnic in He then joined STMicroeectronics as a designer of digita muti-standard decoders for anaogue TV. In 2000 he moved to the company s System Technoogy abs to work on rea-time hardware prototyping for video agorithms. He has aso worked on non-standard sti picture / bayer-pattern compression and on Mutipe Description video coding, and has pubished severa papers on these topics. He hods patents granted in Europe and the USA in digita video processing, sti picture compression, digita moduators and siicon sensors (automotive market). Mr Vitai is now working in the fied of robust source coding, joint source channe coding, adaptive mutimedia pay-out, metadata for mutimedia signas, and graphica interfaces. He gave ectures on Digita Eectronics at Pavia Poytechnic in 2002 and, since 2004, has aso been an externa professor at Bergamo University, Information Science department, where he is teaching Microeectronics EBU TECNICAL REVIEW 43

44 VIDEO STREAMING which may not be avaiabe in the network. Robust source coding expoits the resiience that can be embedded in the bitstream by tuning coding parameters; however it is very difficut to optimize. Mutipe Description Coding aows scaabiity (transmit or decode as many descriptions as possibe), does not require prioritisation, it is very robust (it is unikey to ose a descriptions) and has no a-or-nothing behaviour (decoded descriptions a contribute to decoded video quaity). A standard-compatibe Mutipe Description Coding scheme has been presented: descriptions are created by spatia downsamping in a preprocessing stage prior encoding, they are merged after decoding in a postprocessing stage. MDC performance has been compared to standard coding protected by state-of-the-art FEC: peak quaity of decoded video is ower but it is much more stabe (absence of ciff effect). Severa foreseen appications have been isted, incuding appications in peer-to-peer networks. References [1] F. Kozamernik: Webcasting the broadcasters perspective EBU Technica Review No. 282, March 2000 [2] F. Kozamernik: Media Streaming over the Internet an overview of deivery technoogies EBU Technica Review No. 292, October [3] V.K. Goya: Mutipe Description Coding: Compression Meets the Network IEEE Signa Processing Magazine, September [4] N. Franchi, M. Fumagai, R. Lancini and S. Tubaro: Mutipe Description Video Coding for Scaabe and Robust Transmission over IP PV conference 2003 [5] R. Bernardini, M. Durigon, R. Rinado and A. Vitai: Comparison between mutipe description and singe description video coding with forward error correction MSP [6] A. Vitai, M. Fumagai, draftvitai-ietf-avt-mdc-c-00.txt: Standard-compatibe Mutipe- Description Coding (MDC) and Layered Coding (LC) of Audio / Video Streams Juy [7] S. Astrup and T. Rauhe: Introducing Octoshape a new technoogy for streaming over the Internet EBU Technica Review No. 303, Juy [8] A. Vitai, A. Borneo, M. Fumagai and R. Rinado: Video over IP using standard-compatibe Mutipe Description Coding: an IETF proposa PV conference EBU First pubished : October EBU TECNICAL REVIEW 2007

45 QUALITY OF SERVICE Network structures the internet, IPTV and QoE Jeff Godberg and Thomas Kernen Cisco Systems How woud a broadcaster transmit TV transported over IP packets rather than using traditiona broadcast methods? This artice introduces a view of a generic Service Provider IP distribution system incuding DVB s IP standard; a comparison of Internet and managed Service Provider IP video distribution; how a broadcaster can inject TV programming into the Internet and, finay, how to contro the Quaity of Experience of video in an IP network. Transport of broadcast TV services over Service Provider managed IP networks Access network which takes the data to the home together with the home gateway and the user s set-top box (STB). The whoe network, however, is controed, managed and maintained by a singe Service Provider (SP) which aows him to contro a the requirements The architecture of IP networks for the deivery of inear broadcast TV services ooks simiar to some traditiona deivery networks, being a type of secondary distribution network. The major components are: Super Head-End (SHE) where feeds are acquired and ingested; Core transport network where IP packets route from one pace to another; Video Hub Office (VHO) where the video servers reside; Live Broadcast & VoD Asset Distribution Super Head End IP/MPLS Core VoD Servers IRT/RTE Video Hub VoD Servers Office IRT/RTE Aggregation Network Video Serving Office Home gateway Home gateway Home gateway Home gateway Home gateway Home gateway Video Serving Office (VSO) where access network eements such as the DSLAMs are aggregated; Loca Broadcast Insertion Figure 1 Broadcast TV over an SP-managed IP network 2007 EBU TECNICAL REVIEW 45

46 QUALITY OF SERVICE Thomas Kernen is a Consuting Engineer working for Cisco Systems Centra Consuting Group in Europe. He works on Video-over-IP with broadcasters, teecoms operators and IPTV Service Providers, defining the architectures and video transmission soutions. Before working for Cisco, he spent ten years with different teecoms operators, incuding three years with an FTTH tripe pay operator, for which he deveoped their IPTV architecture. Mr Kernen is a member of the IEEE, SMPTE and active in the AVC group within the DVB Forum. Jeff Godberg is a Technica Leader working for a Chief Technoogy Officer within Cisco. He has been working on IPTV, IP STB design and home networking since 1999, and has been working for Cisco since He was part of the founding group of DVB-IPI and has been working on it ever since, particuary on the home networking, reiabiity, Quaity of Service and remote management parts. Before working for Cisco he designed handhed devices and PC software. needed to deiver a reiabe service to the end point. These requirements are, for exampe, IP Quaity of Service (QoS), bandwidth provisioning, faiover paths and routing management. It is this management and contro of service that separates a managed Service Provider IP deivery of video streams transported over the pubic Internet. The Service Provider acquires the video source in mutipe ways, some of which are the same as in other markets, such as DVB-S. This resuts in significant overhead as the DVB-S/S2/T/C IRDs and SDI handoffs from the broadcasters form a arge part of the acquisition setup. It is therefore preferabe to acquire content directy from another managed network using IP to the head-end, something that is more efficient and becoming more common. Once the content has been acquired, descrambed and re-encoded, it is then carried as MPEG-2 Transport Streams (TS) encapsuated into IP packets instead of the traditiona ASI. The individua muticast groups act as sources for the services which are then routed over the infrastructure, though in some highy secure cases, these may go through IPaware buk scrambers to provide content protection. If security is important, then routers at the edge of the SHE wi provide IP address and muticast group transation to hep isoate the head-end from the IP/ MPLS core transport network. The core network ies at the centre of transporting the stream to its destination but it is the recent deveopments of high speed interfaces that have made it possibe. The ow cost and widey avaiabe Gigabit Ethernet, the more expensive 10 Gigabit Ethernet and the swift 40 Gigabit interface now provide the abiity for the core to transport both contribution and distribution video streams. The modern optics used in these interfaces deiver Bit Error Rates (BERs) and atency that is ower than those of traditiona transports such as sateite. These advantages, combined with an appication ayer Forward Error Correction (FEC) scheme such as the Pro-MPEG Forum Pro-MPEG Code of Practice 4 (CoP4) and IP/MPLS Traffic Engineering (TE) aow for redundant paths across the transport infrastructure. These paths can be designed in such a way that the data fows without ever crossing the same node or ink between two end points, and deivers seamess faiover between sources if the video equipment permits it. In addition, Fast Re-Route (FRR) and Fast Convergence (FC) reduce the network re-convergence time if a node or ink fais to aow for swift recovery, shoud a path fai. The transport stream can aso use the characteristics of any IP network to optimize the path and bandwidth usage. One of these characteristics is the abiity of an IP network to optimay send the same content to mutipe nodes using IP Muticast, in a simiar manner to a broadcast network. This characteristic has many appications and has proven itsef over a ong time in the financia industry, where rea-time data feeds that are highy sensitive to propagation deays are buit upon IP muticast. It aso aows monitoring and supervision equipment to join any of the muticast groups and provide in-ine anaysis of the streams, both at the IP and Transport Stream eve. These devices can be distributed across the network in order to provide mutipe measurement points for enriched anaysis of service performance. The Video Hub Office (VHO) can act as a backup or a regiona content insertion point but aso may be used to source streams into the transport network. This sourcing can be done because of a nove muticast mechanism caed IP Anycast, which enabes mutipe sources to be viewed by the STB as one singe and unique source, using the network to determine source prioritization and aowing for source faiover without the need of reconfiguration. 46 EBU TECNICAL REVIEW 2007

47 QUALITY OF SERVICE Primary and secondary distribution over IP The bandwidth of individua or coective services in primary distribution between a studio or a payout centre and the secondary distribution hubs is traditionay imited by the avaiabiity and cost of bandwidth from circuits such as DS-3 (45 Mbit/s) or STM-1 (155 Mbit/s). This has restricted the deivery of higher bitrate services to such hubs that may benefit from a ess compressed source. The fexibiity of IP and Ethernet removes these imitations and enabes services to be deivered using ower compression and/or with added services. This means that deivery over an IP infrastructure is now possibe: to earth stations for sateite (DVB-S/ S2) based services; IPTV (DVB-IPI) or cabe (DVB-C) head-ends; terrestria (DVB-T) or handhed (DVB-H) transmitting stations. We sha now ook at two exampes of this: firsty, Cabe distribution and, secondy, IP distribution via DVB s IPI standard. Exampe 1: Cabe distribution Cabe distribution typicay foows a simiar pattern to primary and secondary distribution, with the major exception being the use of coaxia cabe over the ast mie. IP as a transport for secondary distribution in systems such as DVB-C has aready been depoyed on a arge scae by different networks around the gobe. Mutipe Transport Streams (MPTS) are run as muticast groups to the edge of the aggregation network where edge QAMs receive the IP services and moduate them onto RF carriers for deivery to cabe STBs. The moduation onto RF carriers can be done in one of two ways: by transating a digita broadcast channe to the STB or by using a cabe modem buit into the STB to deiver it directy over IP. In the atter case, as it is a true IP system, the distribution coud use DVB IPI described previousy without any modification. Today, amost a of the STBs have no cabe modem internay so the IP stream terminates in the hub-site cosest to the STB and even if they did, the data infrastructure is often separate from the video infrastructure. This separation is beginning to change as cabe data modems become much cheaper and the data infrastructure costs become ower. An in-between stage is emerging where most of the broadcast channes are as before, but some of the itteused channes are sent via IP, known as Switched Digita Video (SDV). The consumer notices itte difference between a Switched Digita Video Contribution Network Distribution Network Access Networks Home Network channe and a standard digita cabe channe since the servers and QAMs in the hub and/or regiona head-ends do a the work. The SDV servers respond to channe-change requests from subscriber STBs, command QAM devices to join the required IP muticast groups to access the content, and provide the STBs with tuning information to satisfy the requests. The contro path for SDV requests from the STB is over DOCSIS (DSG), or aternativey over the DAVIC/ QPSK path. In some designs, encryption for SDV can aso take pace at the hub in a buk-encryptor, so minimizing edge- QAM encryption-key processing and thus speeding up the channe-change process. Exampe 2: IP distribution to the STB via DVB IPI DVB has had a technica ad-hoc committee (TM-IPI) devoted to IP distribution to the STB since 2000 with a remit to provide a standard for the IP interface connected to the STB. In contrast to other standards bodies and traditiona broadcast methodoogy, it is starting at the STB and working outwards. In the time since TM-IPI started, many groups around the word have discovered IP and decided to standardize it (see Fig. 2). The standards bodies shown are: DLNA (Digita Living Network Aiance) for the home network see aso the section The Home Network and IP Video ; Studio Mobie Studio IP Studio Fixed Studio Primary Fina Head Studio End Backbone Secondary Head Ends IP IP Home Gateway IP STB STB STB HGI (The Home Gateway Initiative) for the standards surrounding the residentia gateway between the broadband connection and the inhome network; ISMA (The Internet Streaming Media Aiance) for the transmission of AVC video over IP; Figure 2 IPTV-reated activities of seected standardization bodies DSL Forum for the standards surrounding DSL and remote management of in-home devices incuding STBs and residentia gateways; 2007 EBU TECNICAL REVIEW 47

48 QUALITY OF SERVICE (opt) Network Provisioning Server Time Servers DNS Servers DHCP Servers IP Network SD&S SD&S Servers CoD Live SD&S Servers Servers Media CoD Servers Servers Servers Live CoD Media Servers Servers Live Media Servers Figure 3 DVB-IP version 1.3 Architecture ITU which, via the IPTV Focus Group, is standardizing the distribution and access network architecture; ETSI which, via the NGN initiative, is standardizing the IP network carrying the IPTV; ATIS which, via the ATIS IPTV Interoperabiity Forum (ATIS-IIF), is standardizing the end-to-end IPTV architecture incuding contribution and distribution. Nevertheess, the DVB-IPI standard does mandate some requirements on the endto-end system (see Fig. 3), incuding: The transmission of an MPEG-2 Transport Stream over either RTP/ UDP or over direct UDP. The method of direct UDP was introduced in the version of the handbook. Previous versions ony used RTP, and the use of AL-FEC requires the use of RTP. Service Discovery and Seection either using existing DVB System Information, or an a-ip method such as the Broadband Content Guide. Contro of content on demand using the RTSP protoco. DNG Home Network HNED Key: HNED Home Network End Device (e.g. STB) DNG Deivery Network Gateway (e.g. Modem) CoD Content on Demand (e.g. Video on Demand) SD&S Service Discovery and Seection The use of DHCP to communicate some parameters such as network time, DNS servers etc. to the STB. It is norma in IPI to use singeprogramme transport streams (SPTS) as the content are normay individuay encoded and not mutipexed into MPTS as they woud be for other distribution networks. This provides the added fexibiity of ony sending the specificay-requested channe to the end user, which is important when the access network is a 4 Mbit/s DSL network as it reduces bandwidth usage. IPTV and Internet TV convergence The two words of managed STB and unmanaged Internet TV are coming together with sites ike YouTube or MySpace showing user-generated content and excerpts from existing TV programming. Internet TV demonstrates what can be done with an unmanaged network across a diversity of different networks, incuding one in the home. In this section we cover what the home network wi ook ike, compare IPTV to Internet TV, and show how a broadcaster can pace content on the Internet via an Internet Exchange. The Home Network and IP Video Improving technoogies of wireess networks, increases in hard-disk-drive sizes and the increasing number of fatscreen TVs in European househods, makes the home network inevitabe in the near future. Unfortunatey the home network sti remains more of promise than reaity for high-quaity broadcast TV transmission, mainy because the standards and interoperabiity are some way behind. DVB has just reeased a Home Network reference mode which is the first part of a comprehensive specification which wi be competed in The home network consists of severa devices (See Fig. 4): Broadband Gateway Device (BGD) The residentia gateway or modem connected to the IP Service Provider, usuay via either cabe or DSL. Uni-directiona Gateway Device (UGD) A one way device that converts broadcast TV to a stream on the home network. For exampe a DVB-T tuner that converts the stream to IP and sends it wireessy over the home network. Home Network End Device (HNED) The dispay, controing and/or storage device for the streams received either via the BGD or UGD. Home Network Node (HNN) The device, for exampe a switch or Wireess Access Point, that connects the home network together. The Home Network Reference Mode, avaiabe as a separate DVB Bue Book, is based on work done by the DLNA (Digita Living Network Aiance). DLNA aready has existing devices that do stream video over the home network but from sources within the network. The DVB Home Network is the first that integrates both programming from broadcast TV and in-home generated video. 48 EBU TECNICAL REVIEW 2007

49 QUALITY OF SERVICE Externa uni-directiona access Uni-directiona Network (e.g. Network DVB-S) (e.g. DVB-S) UGD exampes: Sateite receivers Terrestria receivers RD exampes: Mobie phone UGD HNED RD ID UGD Comparison of Internet video and IPTV Athough IPTV and Internet-based video services share the same underying protoco (IP), don t et that deceive you: distribution and management of those services are very different. In an IPTV environment, the SP has a fu contro over the components that are used to deiver the services to the consumer. This incudes the abiity to engineer the network s quaity and reiabiity; the bitrate and codec used by the encoder to work best with the imited number of individuay managed STBs; the abiity to simpify and test the home network components for reiabiity and quaity; and prevention of unnecessary wastage of bandwidth, for exampe by enabing end-to-end IP Muticast. HNED Home Network BGD Figure 4 DVB-IPI Home Network Reference Architecture NN HNN BGD HNED Externa Bi-directiona access Externa Network Bi-directiona (e.g. Network Internet) (e.g. Internet) HNED exampes: IP STB, PVR IP STB + PVR NAS, HNED HNED HNN exampes: Switch Access Point BGD exampes: Residentia gateway Broadband modem Contro over the deivery mode doesn t exist with Internet video services. For exampe, IP Muticast depoyments on the Internet are sti very imited, mosty to research and academic networks. This means that Internet-streamed content services use either simpe unicast-based streams between a given source and destination or a Peer-to-Peer (P2P) mode which wi send and receive data from mutipe sources at the same time. One of the other main differences is the contro of the required bandwidth for the deivery of the service. A Service Provider contros the bitrate and manages the QoS required to deiver the service, which aows it to contro the buffering needed in an STB to ensure the audio and video decoders don t overrun or underrun, resuting in artefacts being shown to the end user. Internet video cannot contro the bitrate so it must compensate by impementing deeper buffers in the receiver or attempting to request data from the cosest and east congested servers or nodes, to reduce atency and packet oss. In the peer-topeer mode, ack of avaiabe bandwidth from the different nodes, due to imited upstream bandwidth to the Internet, enforces the need for arger and more distant supernodes to compensate which, overa, makes the possibiity of packet oss higher so increasing the chance of a video artefact. The decoding devices in the uncontroed environment of Internet TV aso imit encoding efficiency. The extremey diverse hardware and software in use to receive Internet video services tend to imit the commonaities between them. H.264, which is a highy efficient codec but does require appropriate hardware and/or software resources for decoding, is not ubiquitous in today s depoyed environment. MPEG-2 video and Adobe Fash tend to be the main video payers that are in use, neither being abe to provide the same picture quaity at the equivaent bitrates to H.264. Chaenges of integration with Internet Video services Internet Video services are growing very fast. The diversity of the content on offer, the ease of adding new content and the speed with which new services can be added is quite a chaenge for managed IPTV services. This eads to the managed IPTV service providers wanting to combine the two types of IP services on the same STB. The most natura combination is the Hybrid mode which has both types of services, probaby by integrating the peerto-peer cient within the SP s STB. This woud aow for coaboration between the two services and woud benefit the users by aowing them to view the Internet video content on a TV rather than a computer. The Service Provider woud then make sure that the Internet video streams obtain the required bandwidth within the network, perhaps even hosting nodes or caching content within the Service Provider network to improve 2007 EBU TECNICAL REVIEW 49

50 QUALITY OF SERVICE deivery. They may even transcode the Internet TV content to provide a higher quaity service that differentiates itsef from the Internet version. This Hybrid mode offers coaboration but may sti incur some imitations. The Internet TV services might be abe to be deivered to the STB but the amount of memory, processing and increased software compexity might make it too difficut within the existing STB designs. This woud increase the cost of the unit and therefore impact the business modes, whist competition between such services may ock out specific payers from this market due to excusive deas. How can a broadcaster get content into an Internet Video service? First some Internet history: Today, the Internet is known wordwide as a magica way to send e-mais, videos and other critica data to anywhere in the word. This magic is not reay magic at a, but some briiant engineering based on a network of individua networks, so aowing the Internet to scae over a period of time to cover the entire word, and continue to grow. This network of networks is actuay a mesh of administrativey independent networks that are interconnected directy or indirecty across a packet switching network based on a protoco (IP) that was invented for this purpose. The Internet mode of a network of networks with everyone connected to everyone individuay was fine unti the cost and size of bandwidth became too high, and the management of individua inks became too difficut. This started the movement towards Internet Exchange Points (IXP) which minimized connections and traffic going across mutipe points by aowing the Service Providers to connect to a centra point rather than individuay connecting to each other. One of the first was at MAE-East in Tyson s Corner in Virginia, USA, but today they exist across Europe with LINX in London, AMS-IX in Amsterdam and DE-CIX in Frankfurt being among the argest and most estabished ones. The Internet Exchange Point, by interconnecting directy with other networks, means that data between those networks has no need to transit via their upstream SPs. Depending on the voume and destinations, this resuts in reduced atency and jitter between two end points, reducing the cost of the transit traffic, and ensuring that traffic stays as oca as possibe. It aso estabishes a direct administrative and mutua support reationship between the parties, which can have better contro over the traffic being exchanged. Being at the centre of the exchange traffic means that IXPs can aow deivery of other services directy over the IXP or across private back-to-back connections between the networks. Today, this is how many Voice-over-IP and private IP-based data feeds are exchanged. This aso makes the IXP an idea pace for Broadcasters to use such faciities to estabish reationships with SPs to deiver inear or non-inear broadcast services to their end users. The independence of the IXP from the Service Provider aso aows content aggregation, whoesae or white-abeed services, to be deveoped and deivered via the IXP. For exampe, Highights of the main areas A/V Encoding FEC Live Broadcast & VoD Asset Distribution MW Servers EPG info quaity GUI design A/V Encoding Super Head End IP/MPLS Core Network Eements Deay, jitter, packet ordering Figure 5 IPTV QoE in the end-to-end mode IRT/RTE VoD Servers District Offices VoD Servers IRT/RTE the BBC in coaboration with ITV is deivering a broadcast TV channe ineup to the main broadband SPs in the UK. They aso provide such a service for radio in coaboration with Virgin Radio, EMAP and GCA. This service has been running for a coupe of years and has been shortisted for an IBC 2007 Award within the Innovative appication of technoogy in content deivery category. Quaity of Experience The Quaity of Experience (QoE), as defined by ETSI TISPAN TR , is the user-perceived experience of what is being presented by a communication service or appication user interface. This is highy subjective and takes into accounts many different factors beyond the quaity of the service, such as service pricing, viewing environment, stress eve and so on. In an IP network, given the diversity and mutipicity of the network, this is more difficut and therefore more critica to success than in other transports (see Fig. 5). Subjective and Objective requirements Subjective measurement systems, such as ITU-R BT , provide a detaied Fast Channe Change, RSVP CAC Centra/End Offices Metro Aggregation Network VOD Server oad distribution Centra/End Office STB A/V decode buffers, Lip sync, Output interfaces Home gateway Home gateway Home gateway Home gateway Home gateway Home gateway Home networking Deay, jitter, packet ordering 50 EBU TECNICAL REVIEW 2007

51 QUALITY OF SERVICE Abbreviations AL Appiation Layer ASI Asynchronous Seria Interface ATIS Aiance for Teecommunications Industry Soutions (USA) AVC (MPEG-4) Advanced Video Coding BER Bit Error Rate BGD Broadband Gateway Device CBR Constant Bit-Rate CoP4 (Pro-MPEG) Code of Practice 4 DAVIC Digita Audio-Visua Counci DHCP Dynamic Host Configuration Protoco DLNA Digita Living Network Aiance home DNS Domain Name System DSG (CabeLabs) DOCSIS Settop Gateway DSL Digita Subscriber Line DVB Digita Video Broadcasting DVB-C DVB Cabe DVB-H DVB Handhed DVB-S DVB Sateite DVB-S2 DVB Sateite, version 2 DVB-T DVB Terrestria ETSI European Teecommunication Standards Institute queryform.asp FC Fast Convergence FEC Forward Error Correction FRR Fast Re-Route GUI Graphica User Interface HGI Home Gateway Initiative homegatewayinitiative. org/ HNED Home Network End Device HNN Home Network Node IP Internet Protoco IPI IPTV IRD ISMA ITU IXP MDI MLR MPLS MPTS NGN NMS QAM QoE QoS QPSK RF RSVP RTP RTSP SDI SDV SHE SP SPTS STB TE TS UDP UGD VBR VHO VoD Internet Protoco Infrastructure Internet Protoco Teevison Integrated Receiver/ Decoder Internet Streaming Media Aiance Internationa Teecommunication Union Internet exchange Point Media Deivery Index Media Loss Rate Muti Protoco Labe Switching Muti Programme Transport Stream Next Generation Network Network Management System Quadrature Ampitude Moduation Quaity of Experience Quaity of Service Quadrature (Quaternary) Phase-Shift Keying Radio-Frequency ReSource reservation Protoco Rea-time Transport Protoco Rea-Time Streaming Protoco Seria Digita Interface Switched Digita Video Super Head End Service Provider Singe Programme Transport Stream Set-Top Box Traffic Engineering (MPEG) Transport Stream User Datagram Protoco Uni-directiona Gateway Device Variabe Bit-Rate Video Hub Office Video-on-Demand mode for picture-quaity assessment by getting a pane of non-expert viewers to compare video sequences and rate them on a given scae. This requires considerabe resources to set up and perform the testing, so it tends to be used for comparing video codecs, bitrates, resoutions and encoder performances. An IP network operator cannot have a team of humans sitting ooking at pictures to assess picture quaity, particuary with the number of channes these days. They therefore test quaity with automated measurement systems which provide rea-time monitoring and reporting within the network and services infrastructure. The measurement systems usuay use some subjective human input to correate a baseine that objective measurement methods can be mapped to. An operator usuay depoys probes at critica points in the network which report back to the Network Management System (NMS) a set of metrics that wi trigger aarms based on predefined threshods. When compared to a traditiona broadcast environment, video services transported over an IP infrastructure introduce extra monitoring requirements. The two main categories of requirements are: IP transport network Whist transporting the services, IP packets wi cross mutipe nodes in the network(s) possiby subjected to packet deay, jitter, reordering and oss. Video transport stream (MPEG-2 TS) Traditiona TS-monitoring soutions must aso be used to ensure the TS packets are free of errors. The two categories are aso usuay in different departments: the IP transport monitoring is within the Network Operations Centre, and the video transport stream monitoring within the TV distribution centre. One of the keys to a good Quaity of Experience in IP is sometimes just good communication and troubeshooting across the different departments EBU TECNICAL REVIEW 51

52 QUALITY OF SERVICE Finay, athough this is beyond the scope of network-based management, additiona measurements shoud be taken into account in a fu system, such as the foowing: Transactiona GUI and channe change response time, service reiabiity. Payoad (A/V compression) Compression standards compiance, coding artefacts. Dispay (A/V decoding) Coour space conversion, de-bocking, deinteracing, scaing. Measurement methods The main measurement methodoogy for the IP transport network is the Media Deivery Index (MDI) as defined in IETF RFC MDI is broken down into two sub-components: Deay Factor (DF) and Media Loss Rate (MLR) which are both measured over a sampe period of one second. The notation for the index is DF:MLR. DF determines the jitter introduced by the inter-arriva time between packets. This shoudn t be viewed as an absoute vaue but is reative to a measurement at a given point in the network. Jitter can be introduced at different points by encoders, mutipexers, buk scrambers, network nodes or other devices. It is important to know what the expected DF vaue shoud be, which can be determined by a baseine measurement in idea operating conditions. The vaue can change dependent on the stream type: Constant Bitrate (CBR) streams shoud have a fixed inter-arriva time whist Variabe Bitrate (VBR) streams wi have a varying vaue. Once a baseine vaue has been determined, you normay set a trigger significanty above this vaue before aerting via an aarm. MLR provides the number of TS packets ost within a sampe period. This is achieved by monitoring the Continuity Counters within the TS. If the stream contains an RTP header, the sequence number can be used for identifying out-of-sequence or missing packets without the need to examine the IP packet payoad. This wi reduce the computationa requirements and speed up the monitoring process. It is norma therefore to distribute MDI probes across the IP forwarding path to aow supervision on a hop-per-hop basis. This heps troubeshoot potentia issues introduced by a specific network eement. To compement the IP packet metrics, DVB-M ETSI TR (ETR 290) is used to provide insight within the transport stream itsef. This operates in the same way as in a traditiona ASIbased infrastructure. The combination of MDI and ETR 290 deivers a scaabe and cost-effective method for identifying transportreated issues. By triggering aarms at the IP and TS eve, these can be aggregated and correated within the NMS to produce a precise reporting too between different events and their insertion point within the network infrastructure. Improving QoE with FEC and retransmission DVB has considerabe experience in error-correction and conceament schemes for various environments, so it was natura given the difficuty of deivering video over DSL that the IPI ad-hoc group shoud work in this area. They spent a significant t i m e c o n s i d e r i n g a a s p e c t s o f error protection, incuding detaied simuations of various forward error correction (FEC) schemes and quaity of experience (QoE) requirements. 2 1 Sends Message to RET server 3 STB Detects Packet Loss STB 4 DSLAM RET server Re-transmits Missing Packet (Muticast Stream ONLY) Assumes Primary Source Figure 6 IPTV QoE in the end-to-end mode 52 EBU TECNICAL REVIEW 2007

53 QUALITY OF SERVICE The resut is an optiona ayered protoco, based on a combination of two FEC codes a base ayer and one or more optiona enhancement ayers. The base ayer is a simpe packet-based intereaved XOR parity code based on Pro-MPEG COP3 (otherwise known as SMPTE standard via the Video Services Forum, see activities.shtm) and the enhancement ayer is based on Digita Fountain s Raptor FEC code ( com). It aows for simutaneous support of the two FEC codes which are combined at the receiver to achieve error correction performance better than a singe code aone. FEC has been used successfuy in many instances; however, another technique in IP can aso be used to repair errors: RTP retransmission. This works via the sequence counter that is in every RTP header that is added to each IP packet of the video stream. The STB counts the sequence counter and if it finds one or more missing then it sends a message to the retransmission server which repies with the missing packets. If it is a muticast stream that needs to be retransmitted then the retransmission server must cache a few seconds of the stream in order to send the retransmitted packets (see Fig. 6). Bandwidth reservation per session One of the advantages of IP is the abiity to offer content on demand, for exampe Video on Demand (VoD). This is resuting in a change in consumer behaviour: from watching inear broadcasts to viewing unschedued content, thus forcing a change in network traffic. This makes corresponding demands on the IP infrastructure as the number of concurrent streams across the managed IPTV infrastructure can vary from thousands to hundreds of thousands of concurrent streams. These streams wi have different bandwidth requirements and ifetime, dependent on the nature of the content which is being transported between the source streamers paying out the session, across the network infrastructure to the STB. The argest requirement is to prevent packet oss due to congestion, which can be prevented if the network is made aware of these sessions and makes sure enough bandwidth is avaiabe whenever setting up a new stream. If there isn t enough bandwidth, then the network must prevent the creation of new streams otherwise a the connected users aong that path wi have a degraded viewing experience (Fig. 7). RSVP CAC (based on RFC2205, updated by RFC2750, RFC3936 and RFC4495) aows for per-session bandwidth reservation to be estabished across the data path that wi carry a given session. Step 1 & 2 in Fig. 7 show the VoD session starting between the STB and the middeware. The authorization credentias wi be checked to make sure that the customer can pay the content, based on a set of criteria such as credit, content rating, geography and reease dates. Once these operations are authorized by the middeware and biing system, the middeware or VoD system manager identifies the VoD streaming server for this session. In step 3, the server initiates a request for an RSVP reservation path between the two end points across the RSVP-aware network infrastructure. Finay, in step 4, if the bandwidth is avaiabe then the session can be initiated; otherwise a negative response wi be sent to the middeware to provide a customized response to the customer. Concusions Deivery by IP of broadcast-quaity video is here today and is being impemented by many broadcasters around the word. The nature of IP as a connectioness and nondeterministic transport mechanism makes panning, architecting and managing the network appropriatey, which can be done with carefu appication of we-known IP engineering. When the IP network is the wider Internet, the ack of overa contro makes guaranteed broadcast-eve quaity difficut to obtain, whereas on a managed IP network, Quaity of Service techniques, monitoring and redundancy can be used to ensure broadcast-eve quaity and reiabiity. The techniques to monitor video are simiar to the ones used for any MPEG-2 transport stream. However, these need to be reated to the IP ayer, for exampe using MDI, as debugging the probem wi often require both network and video diagnostics. EBU First pubished : October 2007 Video on Demand 2 VoD Request Poicy Server RSVP-CAC 1 VoD Request Request Denied/Accepted Avaiabe Bandwidth Check 4 3 Avaiabe Bandwidth Check VoD Servers Figure 7 Connection Admission Contro 2007 EBU TECNICAL REVIEW 53

54 The EBU Technica Review is pubished on-ine, in Engish, four times a year review technica