BBC R&D Annual Review...

1BBC RESEARCH & INNOVATION ANNUAL REVIEW

Preparing the BBC for

Creative Futures and

Digital Britain.

Ensuring it has the

technology to remain

relevant to licence fee

payers and their

evolving needs.

2 APRIL 2006 – MARCH 2007 3BBC RESEARCH & INNOVATION ANNUAL REVIEW

8

10

13

Interactive Television for the On-demand World

Online, On-demand, and On The Move

Freeview Playback

16

19

20

23

Production Magic

D-3 Videotape Preservation System

Programme Production

Radio Spectrum for Production

Contents

PRODUCTION WORKING WITH US

CO

NT

EN

TS

6High Definition Television

14Serving our Audiences – User-centred Research

44

45

The Innovation Forum

Future Media Innovation

47Index

46More Information

NEW SERVICES

24

26

27

28

Digital TV – Switchover

Digital TV – Architectures

Audience Research

DAB and Digital Radio Mondiale

29Kamaelia

DELIVERY

30

32

34

36

Networks for Programme Production

Radio Systems

Video Compression – Dirac and Dirac Pro

Audio Compression

37Digital Rights Management

CORE TECHNOLOGIES

4Foreword

48Credits and address

38

40

43

Collaborative Projects

Standards

BBC Information & Archives


Foreword by HHuuww WWiilllliiaammss,, Head of Research & Innovation

FO

RE

WO

RD

Our achievementsI would like to pick out some of our significant achievementsduring the past year.

Encouraged by the increasing sales of HD-ready televisions,we helped the BBC prepare for the BBC’s HD test trans-missions, which began in summer 2006 on the three maindigital platforms, terrestrial, satellite, and cable.While we awaita decision on the BBC’s launch of a service, we continue tohelp fine-tune the system, to ensure that the public sees thebest possible results from this improved technology.

It is now nearly nine years since we helped to put public digitalTV broadcasts on air in the UK (not forgetting DAB three yearsearlier), and later this year the Digital Switchover will begin inearnest. Although the BBC’s spectrum planners themselves areno longer part of the department, we continue to advise onmethods of technical assistance for our viewers, in particularthe more vulnerable groups.We also work to ensure the BBCcontinues to make efficient use of its radio spectrum allocations,both for its broadcasts and its internal communications.

One of the themes that run through our work is that ouraudiences no longer simply restrict themselves to our lineartelevision and radio broadcasts. In fact some rarely now watchor listen in this way. We have contributed to new developmentssuch as iPlayer, new forms of interactive content on our digitalbroadcasts, Freeview Playback and other ways for our audiencesto control their own viewing. While the PC and 3G phonemight be obvious alternatives to the TV, another excitingprospect is using games consoles to create new forms ofinteractive story telling.

We continue to support our colleagues in production, withinnovative solutions tomake routine tasks easier and to encouragecreative programming.The Piero sports graphics system won theaward for Innovation in Content Creation at IBC 2006.

It is worth remembering that none of this technology will workreliably for consumers, or for the electronic media industryas a whole, unless the right technical standards are in place.We put significant effort into proposing, arguing and supportingstandards, both within the UK and internationally, and valuethe contacts we make through our collaborative work.

GovernanceA new governance structure has been introduced to linkthe work of BBC R&I closely to the business, formalising aninnovation ‘funnel’ to nurture new ideas and research areas andtheir development into products. The link to the business isprovided through regular Technology Executive Conferences,which set strategic technical priorities across the BBC and findsponsors for our projects. A newly created Research Boardapproves and oversees our workplan, deciding which researchshould be continued, and when and where it is appropriateto transfer it to open source or commercial developments.

The year aheadOur work continues, concentrating on eight main researchthemes, with four long term projects that work across them– Digital Switchover, the challenges of the 2012 Olympics,and our academic and industrial collaborations.

I note as we finish writing this Review that the first productsusing our Dirac Pro open-source video compression codec

are about to be shown at the prestigious NAB broadcasters’exhibition and conference in Las Vegas. We are also very closeto trialling the delivery of digital TV and radio channels to 3Gmobiles, bringing the BBC’s programmes to our audienceon the move.

In the coming year BBC R&I will be key to a number of majorBBC projects and initiatives most notably Freesat, iPlayer,further enhancements to Freeview, HD, and helping to defineWeb 2.0 services.

This translates into 14 separate portfolios of research, each withits own portfolio manager:

Production Magic,Wireless Connectivity, Future BroadcastTechnologies, Blue Sky/Future Technologies, External Projects,Programme Production Technology, Interactive TV, InternetDistribution, Digital TV, Metadata Systems, Search & Navigation,Networking & Grid Technologies, Metadata Delivery &Standards, and RF Transmission & Reception.

A new name

This review reflects a year of important

changes, with the renaming of our parent

division to Future Media & Technology (FM&T),

reflecting its role in transforming how the

BBC uses technology. The focus of our work

is moving beyond digital broadcasting to

incorporate the changing world created by

the convergence of broadcasting and the new

media technologies.

We have been joined during the year by the

BBC’s Future Media Innovation and Information

& Archive research teams, whose work is

introduced in this Review.

So we are now BBC Research & Innovation

(BBC R&I).

HHuuww WWiilllliiaammssHead of Research & Innovation


In summer 2006 the BBC began broad-casting high definition television as atechnical trial on digital satellite, onTelewest digital cable, and to a closeduser group on digital terrestrial televisionin the London area.The trial began withcoverage of the World Cup fromGermany. This was rapidly followed byother outside broadcast events includingWimbledon and some of the Proms,together with a range of recorded HDprogrammes shown at peak times.

The trial has posed many challenges, andwe are taking a leading role by providingadvice and assistance to BBC Sport, BBCResources, Red Bee Media and Siemens,and collaborating with equipment manu-facturers. HD technology is still relativelyimmature, and the entire signal chainfrom contribution, through productionand post-production to play-out anddistribution, has to work at much higherbandwidths and bit rates. The referencetest system that we maintain atKingswood Warren for digital televisiondevelopment is crucial. It enables usto give advice based on practicalexperimentation in a controlled andrepeatable environment on matterssuch as picture quality, bit rates, codingparameters, picture/sound timing,

multi-channel audio, and subtitles.We have led the work to edit therelevant DVB standard to enablesubtitles at HD resolutions.

On satellite, the transponder with themost appropriate coverage for the trialwas already full with standard definitionservices. A method had to be devised tomove these into newly acquired satellitecapacity (the BBC’s seventh transponder)without disturbing the service to theviewers. As on previous occasions, thiswas a collaborative effort between us,Siemens and BBC Distribution. At thesame time, the BBC took the opportunityto expand its provision of interactivestreams for a temporary period, allowingfull interactive coverage in standarddefinition of Wimbledon and the WorldCup, in addition to the simultaneous HDcoverage of these events.

For the cable trial, we provided technicalsupport to BBC Distribution indiscussions with Telewest and helped ona number of technical matters such asspecifying interfaces and trouble-shootingat the start of the trial.

For the terrestrial trial, we defined thetechnical requirements for the HD set

top boxes that were to be providedto the trialists, assessing a large numberof submissions from manufacturers inresponse to a tender. The short time-scale, five months, and cost constraintsmeant that only a limited set of featurescould be built into the boxes. Wetherefore issued a list of essentialrequirements, in particular being ableto receive Dolby Digital surround sound,and a longer list of desirable features,for example, reception of standarddefinition Freeview broadcasts.

Boxes from two manufacturers wereshortlisted, on a combination oftechnical capability and price.We helpedthese manufacturers finalise the designof their boxes in time for the trials.

Trialists were required to possessdisplays that conformed to the EICTA‘HD Ready’ standard. Such displaysare fitted with either an HDMI or aDVI connector for the HD signal.Because these are relatively newinterfaces, we carried out compatibilitytests between a selection of typicalHD Ready displays and the trial boxes,finding almost all pairings to becompatible. We also took the oppor-tunity to measure picture overscan,

to recommend a safe visible area forcaptions and other graphics.

Objectionable changes in loudness werefound when switching between programmesusing Dolby Digital and MPEG Layer IIaudio coding. We found this effect wasbeing introduced by the mechanisms forsignalling and modifying audio levels inDolby Digital, and we are now workingwith manufacturers to reduce it.

Appreciation of HDTV has two mainaspects, programme content andtechnical quality. Clearly, our professionalinterest is in the latter! Comprehensivequestionnaires were put out by BBCMarketing, Communications andAudiences, and trialists were also ableto post comments on a private webforum. Generally, the responses werepositive, particularly concerning picturequality, and there was also a call formore programmes with surround sound.Critical comments included an occasional

lack of lip synchronisation, and somepops and clicks on the sound, thesebeing due to teething troubles withthe technical systems.

Over a few days in October 2006 wedeliberately varied two technical para-meters; bit rate, between three differentvalues, and the broadcast standard,between 1080i and 720p. From commentssolicited on the web forum we foundthat some trialists were aware ofreductions in bit rate in both the 1080iand 720p modes. A second longer seriesof tests has just been completed.

The latest upgrade to the video data-compression coder is being tested to assessthe improvement in coding efficiency.

The BBC now needs to ensure that thetechnical requirements of an HD servicecan be met. The trial highlighted anumber of areas where the specificationneeds more work to ensure thatmanufacturers can produce compatiblereceivers. The DTG has started toaddress this, and we are activelycontributing on aspects such as thedetailed syntax for the video encoding,system signalling, and requirements forreceiver video and graphics processing.

NE

W S

ER

VIC

ES

High Definition Television

Many BBC programmes are already made in HD, and the amount

will increase over the next few years as studios and other

equipment are upgraded as part of the normal replacement cycle.

The cost of HD production is becoming closer to SD, and it may

soon be difficult to obtain SD production equipment.The BBC also

has to protect its sales to other broadcasters.This is a valuable

source of income, particularly from North America, where HD

content is now insisted upon. HD production also ensures the

BBC has a reasonable stock of programmes to show on a service,

although a decision on a launch has yet to be taken.

What will replace HD?It is not too early to ask this question.

Even though HD broadcasting has

barely begun, the basic research was

started more than 20 years ago. Will

the future be ultra-HD, 3D-TV, or

something else? While it is clear that

there is a continuing quest for ever

higher quality pictures and sound, as

can be seen from 4k digital cinema,

equivalent to 2048 vertical lines, and

NHK’s 4096-line Super HiVision

system, there are serious difficulties

in delivering this to the home.

Or will ultra-HD become a means to

an end? There are benefits for pro-

gramme production in starting with

such high-quality material. If it is used

correctly, it can help maintain HD

quality at the end of the production

chain, and this encourages more

elaborate processes, such as the

‘Production Magic’ described later.

Currently we are exploring the state

of research worldwide and investi-

gating opportunities for collaboration

with other organisations.

NE

W S

ER

VIC

ES

‘‘GGeenneerraallllyy,, tthhee rreessppoonnsseesswweerree ppoossiittiivvee,, ppaarrttiiccuullaarrllyyccoonncceerrnniinngg ppiiccttuurree qquuaaiilliittyy’’

9BBC RESEARCH & INNOVATION ANNUAL REVIEW8 APRIL 2006 – MARCH 2007

With mass storage available in thereceiver, large amounts of data canbe downloaded, perhaps at low speedovernight or at any other time whenbroadcast capacity is available.The inter-active services can then include richergraphics and even audio-visual clips.

To explore the possibilities, the BBC hasbeen working in partnership with CabotCommunications, a supplier of digitalTV software technologies, to extend thecapabilities of a standard consumer digitalTV recorder. The extensions allow therecorder to download and store MHEG-5applications for on-demand access andto allow these applications to scheduleconventional recordings.

This has allowed the BBC to undertakea three-month technical trial, with300 of the modified recorders issuedto members of the public.The con-sumer proposition for this wasa ‘catch-up TV’ service, where therecorder automatically records aselection of BBC programmes fromthe broadcast schedule.These are thenmade available for on-demand accessvia a rich graphical user interfacewritten in MHEG-5 and enhanced withdownloaded video clips.

The extensions developed to supportthe trial open up exciting possibilities forenhanced television services, such asDVD-style extras providing ‘the makingof...’ or ‘an interview with...’. We havesuccessfully tested some examples ofthis in the laboratory, and are startingto include them in the technical trial.

The BBC is gaining valuable experiencethrough the trial on the practicalities ofrunning this type of service and work isunderway to explore how to standardisethe extensions that enable them.

The number of homes with broadbandInternet connections is growing rapidly,and BBC R&I has been looking at howa receiver with both an aerial input and abroadband connection might combinecontent from both sources. Some initial

ideas developed by the BBC were shownat IBC 2006. The demonstration showedtext, graphics and video delivered overa broadband connection, using unicastand multicast, viewed alongside conven-tional broadcast-delivered content.

However, adding broadband accessto set-top boxes and TVs in an openmarket such as Freeview presents anumber of challenges. Home ADSLinstallations vary from a modemconnected directly to a single PC,to complex networks with firewalls.Connections come with a range ofconnection speeds, often with usagerestrictions and no guarantee of thequality of service. BBC R&I is workingwith the industry on how to deal withthese issues, through the DTGInteraction Channel Working Group.

The majority of the interactive services that the BBC has delivered

to date have been based on content delivered over a unidirectional

broadcast path. This path is shared with the main television pictures

and sound, access services, schedule information and other data,

and this generally limits the interactive content to relatively modest

text and image files. However, two receiver technologies, mass local

storage and broadband connectivity, can not only radically improve

the nature of interactive content but also the way viewers can

access it.

Interactive Television for the On-demand World

The portable content formatThe BBC provides interactive televisionon a number of different digital TVplatforms. This is a challenge becauseeach platform follows a different setof technical standards for interactivity,and the BBC therefore has to produceseveral versions of every piece ofinteractive content.

This problem is not new, and the BBChas been exploring ‘author once, publishto many’ approaches for over five years.The result, in collaboration with otherorganisations, is a new DVB standard,the Portable Content Format (PCF).

The PCF, published by ETSI in January2006, is being embraced by the BBCas both an authoring format and asan input format for platform-specificpublication processes (often calledmore simply, ‘transcoders’). It is alsohoped that PCF can make it easierto take in interactive content fromthird parties, reducing both the ‘timeto air’ and the management costs.

Interactivity at HDLarger displays and greater screenresolution gives HD television a ‘wow’factor that is hard to resist. But what part

do interactive services play alongside HDprogrammes? The BBC is researchingways to broadcast and presentinteractivity efficiently on HD, as far aspossible with an awareness of thetechnology within HD receivers, butperhaps needing to influence it too.

Initially, HD programmes could make useof the interactive services transmitted forSD, such as BBCi and the populartelevision multiscreen services such asWimbledon tennis. But how best shouldan HD receiver reproduce this content?Is it acceptable simply to render it at SDand then re-scale, or would it be betterto have intelligent software rendering itdirectly at HD?

Beyond this some interactive servicescould benefit from being authoredspecifically for HD. This could be asstraightforward as providing images atHD resolution, so that no up-conversionis required, or redesigning to make useof the extra resolution and thelarger screen.

We continue to work with otherinterested parties, to consider the impactof such ideas on the broadcast stream,in terms of need for additional capacity,

and on the receivers, in terms ofmemory and processing demands.

Interoperability on DTTFreeview transmissions conform to a setof published technical standards that theBBC helps to write. This allows anymanufacturer to produce digital terrestrialtelevision receivers for the UK. However,the lack of a single platform operatorand the open market in receiverscontinue to present a conformanceand interoperability challenge, particularlyin the context of interactivity. This canlead to receivers failing to display someinteractive content correctly.

BBC R&I plays a major role in the waythat the BBC resolves this, by workingclosely with individual manufacturersand across the industry, through bodiessuch as the DTG MHEG InteroperabilityGroup.Through this it aims to makethe viewers’ experience more reliableand consistent.

‘‘TThhee eexxtteennssiioonnss ddeevveellooppeedd ttoo ssuuppppoorrtttthhee ttrriiaall ooppeenn uupp eexxcciittiinngg ppoossssiibbiilliittiieessffoorr eennhhaanncceedd tteelleevviissiioonn sseerrvviicceess,, ssuucchhaass DDVVDD--ssttyyllee eexxttrraass pprroovviiddiinngg ““tthheemmaakkiinngg ooff......”” oorr ““aann iinntteerrvviieeww wwiitthh......””..’’

NE

W S

ER

VIC

ES

‘‘BBBBCCRR&&II aaiimmss ttoo mmaakkee tthheevviieewweerr’’ss eexxppeerriieennccee mmoorreerreelliiaabbllee aanndd ccoonnssiisstteenntt’’


television. Colleagues in BBC FM&Tdeveloped a peer-to-peer Internetdownload service for distributing largemedia assets to personal computers forthe ‘iMP’ trial of 2005/2006.

In 2004 we set up a public trial of livemulticast streaming. Multicast uses therouters inside the network to duplicatethe streams of data packets rather thansending separate streams all the wayfrom the head end to each recipient.Such trials promote the roll-out ofmulticast, by encouraging users to setup their home equipment to receiveit, and Internet service providers toset up their networks to carry it.

This trial ran successfully until August2006, at which time the rights we hadnegotiated for distribution of the contentexpired. It is now to be developed intoa full broadband ‘iPlayer’ service, followingthe approval of the BBC Trust.

Research continues on other deliverymethods for on-demand video content,beyond the iPlayer environment andlooking at future home networksdelivering the same content transparentlyto the TV, PC, or mobile devices.

Syndicating on-demand contentThe BBC intends to syndicate on-demand content to cable operators,IPTV service providers, and 3G mobilephone networks, rather than deliveringdirect to these operators’ customers.

Each of these platforms uses a differentcombination of technologies, some ofthem proprietary. BBC R&I has helpedBBC Distribution to design the technicalfacilities for preparing the content forthe various platforms.

The distribution system will be operatedby the BBC’s playout contractor Red BeeMedia. It is envisaged that a commoncontent ingest and media managementsystem called the On-Demand ProductionSystem (ODPS) will be shared with thebroadband iPlayer to avoid needlessduplication of effort. The infrastructurecan acquire live programmes in real timeas well as the more conventional methodof ingesting tapes or files ahead of lineartransmission. A suite of encoders will thenconvert the content into the appropriateformat for each platform.The resultingmedia files will be staged to a ‘neutraldrop-off zone’ operated by our

technology partner Siemens, from wherethe operators will retrieve them.

Metadata is sent direct to the platforms.This metadata will populate the contentguides, and will include extras such asartwork relating to the programmes andpromotional trailers; most on-demandplatforms allow a sequence of promotionalitems to be played automatically beforeand after the programme that the viewerhas selected.We hope to establish aprofile of TV-Anytime for the metadatathat is acceptable to all parties.

10 APRIL 2006 – MARCH 2007

Some broadcasters may view the newtechnologies as a threat to existingbusiness models.The concept ofadvertising ‘spots’ is particularly at riskwhen the viewer can order individualprogrammes or simply skip throughunwanted content.The BBC, on theother hand, is embracing this new non-linear landscape as an opportunity toexploit its content to the full.We canserve our licence fee payers better byallowing them to watch our programmeswhen they want to, not just at the timeswe choose to schedule them.

The BBC has been streaming audio-visualmedia on the Internet for some years.Indeed, the BBC was pivotal in the devel-opment of these types of services.TheRadio Player service, based on technologyoriginally developed by BBC R&I, hasproved popular and demonstrated a clearconsumer demand for ‘listen again’ services.The challenge in the present Royal Charterperiod is to provide an equivalent ‘watchagain’ service for television.

Television is a greater technical challengethan radio.The unicast streamingtechnology used to deliver the RadioPlayer service does not scale to the highbit rates and file sizes demanded by

NE

W S

ER

VIC

ES

Online, On-demand, and On the Move

The end of television delivered as a scheduled broadcast service

has been widely predicted for some years now, with little evidence

to back up the claim. The VCR allows viewers some freedom

from the broadcasters’ programme schedules, but most models

are notoriously difficult to program, generate a stack of tapes,

and will only record one channel at a time.A new breed of

on-demand technologies is finally emerging that we believe will

disrupt the status quo in a more profound way.

NE

W S

ER

VIC

ES

True video-on-demand servicesstream programme content in real timeto a set-top box or personal computervia a broadband connection.The viewerselects programmes from an electroniccatalogue and can start watchingimmediately; unlike near-video-on-demand there is no need to wait untilthe next showing starts.The on-demandreceiver can even be combined with adigital TV recorder to provide the bestof both worlds.

The digital TV recorder is a digitalTV receiver that records onto a highcapacity hard disc drive instead of tape.In addition to the usual VCR functions,programmes can be paused and resumedwhen watching ‘live’, and recordings canbe viewed before they are complete.The digital electronic programme guidesmake recordings easy to set up, and ifsuitable extra signalling is included in

the broadcast, the machine can be setto record entire series just as easily assingle programmes.The recording timecan even adjust itself automatically if theschedule changes. On the latest modelspart of the capacity can be reserved forrecordings selected by the serviceprovider (‘push-video-on-demand’).

Download services offer programmecontent over a broadband connectionto be stored for later viewing. Althoughusually associated with personalcomputers, these services have alsobeen demonstrated with broadband-connected set-top boxes.The contentis encrypted and the viewing permissioncan be set to expire after a fixed period.

The BBC iPlayer is a unified onlineplayer, to find and play on-demand andlive-streamed audio and video.

Convergence of walled gardens and the InternetVideo-on-demand is an emergingbattleground for establishedorganisations to test their resolve inmaintaining old business models whenInternet start-up companies start tooffer similar services; cable companiesand telcos are building ‘walled garden’video-on-demand systems to combatthe pressure from Internet delivery ofa broader, potentially more attractiveset of content.The BBC’s public serviceremit means it is providing content tocarriers in both camps, as well as directdelivery though projects like iPlayer.

‘‘ WWee ccaann sseerrvvee oouurr lliicceennccee ffeeee ppaayyeerrss bbeetttteerrbbyy aalllloowwiinngg tthheemm ttoo wwaattcchh oouurr pprrooggrraammmmeesswwhheenn tthheeyy wwaanntt ttoo......’’

TV and Radioon 3G mobilesEarly in 2007 the BBC was given approvalto launch trials to deliver BBC TV andRadio channels to a number of 3Gmobile operators. The trial will belaunched in April 2007 on Orange,Vodafone and 3, and to meet this tighttimescale it has been necessary to reusethe technical infrastructure alreadydeployed for the multicast trial. This willbe replaced in due course, to allow forexpansion of the trial to more operatorsand to include a wider range of BBCdigital TV services and content,

At first, the trial will be carrying threeBBC TV channels – BBC One, BBCThree, and BBC News 24 – and abouteight Radio channels, all streamed live.It will be receivable on most modern3G phone models, although usage mayincur a charge from the user’s mobilephone operator.

Users will receive the appropriateregional variant of BBC One, dependingon their location when calling in.(The BBC Three and BBC News 24channels have no regional variations.)Some schedule data is carried, to displayprogrammes titles and synopses.

Each video signal is encoded at 100 kbit/s,using H.264 coding.The image is clippedslightly to suit the narrower usable aspectratio of the typical mobile phone screen.The audio is encoded using AMR(adaptive multi-rate coding), which is

already widely used on mobile phones.AMR is however optimised specificallyfor speech, and we hope later to moveto AAC coding to improve the overallsound quality and make a further savingin bit rate.

Transmission is by the conventionalmechanism used for IP streaming tomobile devices. At present a separatestream is set up for each viewer. Themobile phone operators may decideto change this in future, although thenumber of simultaneous calls to the sameTV or radio channel within a singlemobile phone cell may not always justifya different arrangement.

Because any 3G phone that can receivethe service will also have Internet access,it is not essential to embed the inter-active content that accompanies thedigital broadcasts.The same information,and more, can be obtained from the BBCwebsite. Having a mix of content may bea catalyst for the way data services tomobile phones are charged.

12 APRIL 2006 – MARCH 2007 13

NE

W S

ER

VIC

ES

BBC RESEARCH & INNOVATION ANNUAL REVIEW

Freeview Playback is much more than a marketing exercise. It is underpinned bysignificant enhancements to the signalling that accompanies DTT programmes.These enhancements were developed by BBC R&I in consultation with otherbroadcasters, industry, metadata aggregators, and colleagues in BBC Marketing,Communications & Audiences.This wide collaboration was essential because thedata to support Freeview Playback runs through the whole broadcast chain.

An initial phase of the work was to agree an amendment to the UK guidelines forbroadcaster and receiver equipment interoperability, to describe how FreeviewPlayback would work ‘under the hood’.This was coordinated and edited under theauspices of the DTG by BBC R&I, and relied on our long term involvement with otherstandards bodies such as DVB.

To bring Freeview Playback to the viewer substantial changes were needed to theDTT technical systems. All electronic programme information for the UK DTTplatform is generated by a system called the central collator, now owned andoperated by our technology partner Siemens.When this came due for replacementBBC R&I made sure that the new system would be capable of supporting FreeviewPlayback; our involvement in the replacement continued right through to the finalstages of systems integration and testing.The central collator dated from the start ofdigital broadcasting, and it was a major achievement when this broadcast critical systemwas successfully changed over while services remained on-air.

Once the new system was in place the necessary extra metadata had to be generatedand new feeds arranged to supply it. As the acknowledged experts in this area, BBCR&I advised the BBC’s metadata supplier on suitable solutions.The Freeview Playbackmetadata is compliant with the TV-Anytime standard.

The BBC was the first broadcaster to transmit Freeview Playback information.A commitment has been made by the other UK DTT broadcasters to follow suit.This co-operation is encouraging manufacturers to produce the receivers, a marketwe expect to expand during Digital Switchover.The first two models have beenavailable since December 2006.The features provided by Freeview Playback have alsobeen recognised in Scandinavia, where the 'NorDig' broadcasters plan to implementthe same underlying signalling.

The business relationships established through Freeview Playback continue, and areexpected to lead to further developments. Currently, we are working on a new featurecalled ‘trailer booking’, which allows a Freeview Playback viewer to book a recordingwhile watching a trailer for the programme.

Freeview Playback

Freeview Playback brings a set of new features and the mark of

quality assurance to the free-to-air Digital TV Recorder market.

It is the first major change to the Freeview platform since the

introduction of seven-day programme guides. Getting the service

on air at the end of 2006 was the culmination of work that started

for BBC R&I more than two years ago.

• Accurate Recording – a FreeviewPlayback recorder continuallymonitors broadcaster informationto control the start and end of arecording and does not just relyon a pre-published time.

• Series linking – a Freeview Playbackrecorder can automatically identifyand record programmes from thesame series.There is no longerthe need to rely on recording thesame time and channel each weekor to try to match programme titles.

• Repeat identification – when theviewer wants to record programmesthat overlap, the recorder can checkif the clash can be avoided byrecording a repeat.

• Split programmes – this featureaddresses a long term issue ofprogrammes transmitted in multipleparts, for example a film split by thenews.Viewers often fail to recordthe second part of the programme.Freeview Playback ensures nothingis missed.

• Recommended viewing – therecorder can offer the viewer otherprogrammes to record, related toother recordings that havebeen made.

• Compliance – before carrying theFreeview Playback logo, a recordermust pass a rigorous set of tests todemonstrate it complies with thetechnical standards and features.

Online, On-demand, and On the Move continued

‘‘EEaarrllyy iinn 22000077 tthhee BBBBCC wwaass ggiivveennaapppprroovvaall ttoo llaauunncchh ttrriiaallss ttoo ddeevvlliivveerrBBBBCC TTVV aanndd RRaaddiioo cchhaannnneellss ttoo aannuummbbeerr ooff 33GG mmoobbiillee ooppeerraattoorrss..’’


bring the BBC tradition of rich narrativeinto this domain, looking at howto author and deliver truly interactive,open-ended stories.We are taking partin an EU-funded project, CALLAS, whichis exploring combining emotion-sensinginterfaces with interactive narrativetechniques to produce a ‘bardicinteractive storytelling engine’.

This work is highly experimental andin its early stages, and it is difficult topredict what might emerge. Early ideascentre on having a narrator with whomthe viewer or viewers interact byspeaking.The story is driven by theoutcome of these interactions, withthe system working from not onlywhat is said but also how it is said.Virtual actors could be used for thenarrator and the cast, so that actionneed not be pre-recorded. Each vieweror group of viewers will therefore seea different story unfold.

Mass participationOur audience’s ability to send contentto us, as demonstrated by the waves ofmobile phone images we receive whenmajor news stories break, shows thewillingness of our audience to participate.

We are involved in a three yearDTI-funded collaborative projectcalled, very appropriately, ‘Participate’.This project is exploring the convergenceof online and broadcast media to createnew kinds of mass participatory eventswhere a broad cross-section of thepublic contributes and shares content.This might be centred on one of thelarge outdoor displays the BBC operatesfrom time to time, for example.

We are working with BBC Future MediaInnovation and BBC Production to definesome of the narrative aspects of theinitial trials.We are trying out potentialtechnologies to allow large numbers ofpeople to access the system simultane-ously, even when on the move withoutcompromising technical quality.

Behavioural scienceParticipate is a good example of theneed to understand human behaviour –how do people engage with a publicevent such as the one just described?Our immediate challenge in Participateis to decide how to measure this, andwhat metrics one might use.

Behavioural science factors, such asusability, accessibility, user experience and

interaction design are now common andcritical features of the mainstream BBCbusiness. We are continuing to support anumber of projects across the BBC, includingswitchover and new interactive platforms.

Image classificationOpening up the BBC to user-generatedcontent creates a practical problem insorting the large volumes that can besubmitted in a short space of time.The material is often topical and mustbe dealt with quickly.

We have been working on a project toclassify user-generated content to allowrapid selections to be made.This workis a collaboration between ourselves andthe University of Surrey.

We hope that ranking submitted imagesby their similarity to wanted images,clustering related images together, mightachieve quite good results while beingrelatively simple to achieve. A moreelaborate technique, which mightultimately achieve better results, is togenerate metadata by object recognition.Although we started here with mathe-matical methods, we are now exploringtechniques that try to emulate therecognition processes in the human brain.

14 APRIL 2006 – MARCH 2007

Discovering contentContent is becoming easier to access,but harder to find. There are moredevices supporting even more typesof media, and there are more waysto deliver it to the viewer. In addition,viewers are expecting more from media– they want links between related items,and they want to navigate instantly tothe parts that most interest them.

This is changing how we describe andprovide our content. And of courseit’s not just about BBC content – ouraudience are contributing more, not justin terms of images, audio and videobut also opinions, facts and data.

In order to encourage innovation inthe methods of discovering our contentwe are continuing to supply feeds ofelectronic programme data based on theTV-Anytime standard to the developercommunity on BBC Backstage. We havecreated a simple means for developersto query this data via a publicly availableexperimental Web ApplicationProgramming Interface (API).

The Web API allows developers to createthird party tools and ‘widgets’ to displayschedule information for all our channels.

We have been experimenting with howwe can combine the data and storedcontent to provide interesting means offinding content, creating a simple ‘player’that combines live video/audio streamswith programme information in a webpage.

Content packagingThe success of Podcasting – downloadableaudio programmes – suggests thatpeople like to acquire content for laterconsumption. However, the playback isvery basic, with limited opportunities forenhancements or to promote otherservices and content.

The idea behind content packaging is toreplace the download of individual itemswith a personalised content package,based on the profile of the user. Thepresentation of the content is created bythe content provider, in the appropriateformat for the playback device, usinga player or browser application.

We have developed two demonstratorsfor content packaging, the first designedfor mobile phones, the second forhigh-end platforms such as laptops.We are using these demonstratorsto develop further ideas and testtheir usability.

Games consolesOne of the BBC’s primary goals isto reach and serve the entire Britishaudience. A growing group that areseen by many to be under-served bythe BBC are computer games enthusiasts.This group tend to feel that their gameplatform is the primary platform forrelaxation and entertainment, rather thanthe TV; the BBC is felt to be insignificant,if not irrelevant. One way to reach thisaudience is to provide content for theirplatforms of choice, using the contentgrammars that are familiar to them.

As a first step we are assessing ourexisting content, to find what is mostappropriate editorially and technically forgames consoles. We have successfullydemonstrated play-out of live streamedBBC video on Playstation2TM andPlaystation3TM consoles, using an opensource player, and we are now lookingto cover a broader range of consoles.We are also considering how the BBC’sportfolio of web-based games can beadapted for console play.

Interactive story-tellingWith the benefit of this experience withgames consoles we can start to try newconcepts.We are researching how to

NE

W S

ER

VIC

ES

How do we stay in touch with our audiences? How do we ensure

that what we produce is relevant to them? How do we help them

navigate the current maze of media offerings?

This set of projects is pursuing a number of ideas, under the

general heading of user-centred research. It also includes our

behavioural science work, where we try to understand the basic

human needs that drive media behaviour, and to assess their

impact on our audience.

Serving Our Audiences – User-centred Research

NE

W S

ER

VIC

ES


Inserting graphicsinto live imagesWhenever a production team plansto insert a virtual object into an image,the graphics system needs to knowprecisely how the camera is moving, sothat the virtual object can be renderedin the correct place in every frame.Although there are several toolsavailable commercially for doing thisin post-production, live (or as-live)operation is more difficult. For studio-based productions, it is usually necessaryto shoot in a studio equipped witha camera tracking system, such as thefree-d system we developed some yearsago, which uses markers mounted onthe ceiling. For other kinds of production,such as outside broadcasts or shoots onlocation, new solutions are needed.

We have been working with severalEuropean companies and universitiesin the MATRIS project to developmarkerless camera tracking usingnaturally-occurring features in the scene.This three-year project was funded bythe EU’s 6th Framework Programme,and finished in January 2007. In additionto tracking directly from the cameraimages, the project also developed aninertial sensor, and looked at the use

of an auxiliary camera fitted with afish-eye lens. Both of these techniquescan improve the accuracy and stabilityof the tracking. We coordinated theproject demonstration at IBC 2006,which attracted the interest of severalpotential licensees.

As a part of our work in the MATRISproject, we developed a method thatmeasures camera movement by trackingthe lines on a football or rugby pitch.Thishas been licensed to Red Bee Media as a

part of the Piero sports graphics system,and is now in regular use by the BBC anda number of other broadcasters aroundthe world. In June 2006 the system wasused in High Definition for the FootballWorld Cup, and for the first time it useda method to automatically lock onto thepitch lines when starting to track.

In some applications, it is useful for apresenter to be able to pick up a virtualobject, rather than it being fixed in thestudio or on the ground.We previouslydeveloped and licensed a system knownas MixTV, which tracks a specially-patterned card which a presenter canmove, so that a virtual object can berendered on top of it. In April 2006,a version of the system was used in athree-month trial by BBC Jam, the BBC’sbroadband learning service for five to 16year olds. This allowed users to interactwith virtual 3D objects in real time, usingtheir own hands rather than a mouseor a keyboard.

We have been working with BBC Newsto develop a system that tracks thecameras using an array of markerpatterns placed on the walls of thestudio or on other par ts of the set.This avoids the expense and complication

of installing a dedicated tracking system,and can be used in almost anylocation. Another approach beinginvestigated is to display the markerarrays on an existing rear-projectionsystem, so they can easily be switchedoff when vir tual graphics are notrequired. The system could alsobe used in conjunction with hand-heldmarkers to allow both fixed andmoveable virtual graphics to be shown.

iview – capture of 3Dscenes and generationof virtual viewsWe are leading a DTI-funded collaborativeproject called iview, to develop methodsto capture action from events such asa football match in 3D and to provide a360 degree free-viewpoint video replayof the action. During the replay theposition of the virtual camera can bemoved freely and crucial moments ofa game can be seen from viewpointswhich are not normally open to a realcamera. BBC Sport has an obviousapplication for such a system, as it wouldallow them to fly a virtual camera aroundthe pitch, giving a new way of presentingkey moments. It could also generate top-down views for tactical analysis, or viewsas seen by a linesman, player or referee.

In its first year, iview applied techniquesformerly developed for use in a chroma-key equipped studio for use outdoors.New and robust techniques wererequired for calibration of the cameras,keying and generation of the 3D modelsof the action. First tests in collaborationwith BBC Sport and BBC OutsideBroadcasts were carried out for footballusing either a set of specially mountedfixed cameras or just the camerasnormally present. From the fixed cameraset, virtual camera positions (such as thegoal keeper’s view) can be generated toprovide new insights into the game thatare not possible to achieve otherwise.Current work is focussing on tests withHD cameras.

The processing and generation of the3D model of the action is currentlycarried out off-line, but options for real-time processing are being investigated.The replay uses image-based renderingmethods that can be implemented atinteractive rates on standard PCs. Onegoal of iview is to stream the replay datato an interactive platform such as agames console so that the user candetermine his or her preferred viewpoint.We are also looking at applications insports other than football, and in other

programme genres including studio-based programmes.

Tracking peopleand objects in 3DBy tracking people in a scene, it ispossible to automate some productionprocesses. We have been investigatingapplications involving the automaticcontrol of lights, by tracking the positionsof the camera and presenter. The abilityto automatically keep a presenter lit ashe or she moves around a studio offersthe prospect of reducing the overalllevel of lighting needed, whilst ensuringthat a sufficient level of light is used.This is especially useful to satisfy therequirements of HD production wheremaintaining a good depth-of-fieldbecomes particularly important.Workingto requirements from BBC News, wehave demonstrated a prototype systemthat can do this using information fromthe cameras.The system then adjusts thelighting by controlling the direction, beamwidth and brightness of a number ofmotorised spotlights.

We have also been investigating the useof audio location techniques to track theposition of people. We have built areal-time demonstrator that uses two

Production Magic

We are developing a range of production tools that harness the

power of computer image processing to analyse a scene in 3D.

With these tools, we can insert virtual objects into a live scene,

generate a new view of a real scene from a virtual viewpoint,

or automatically steer a camera or spotlight to follow a person

around the set.

This work ranges from relatively short-term projects for immediate

production needs, through to longer-term fundamental research

that could lead to radical changes in the way programmes are made.

It may also lead to new forms of 3D or immersive content that

go well beyond what is possible today.

‘‘ IInn SSeepptteemmbbeerr22000066,, tthhee Pierossyysstteemm wwoonn aann IIBBCCIInnnnoovvaattiioonn AAwwaarrdd,,iinn rreeccooggnniittiioonn ooff iittssrraannggee ooff nnoovveellffeeaattuurreess tthhaatt ccaammeeoouutt ooff BBBBCC RR&&IIwwoorrkk.. ’’

PR

OD

UC

TIO

N

Production Magic continued

pairs of microphones to estimate thelocation of a speaking person, using timedelay analysis. This could be used tolocate and track a TV presenter, ormembers of a studio audience when theyspeak, or in outdoor sporting events tohelp automate camera framing andshot selection.

Although these developments are partof our long-term work to create newways of making programmes, thetechnology we develop along the waycan also be rapidly deployed to addressshort-term problems. One example ofthis was a requirement for a simple wayof gathering votes from an audience.We developed a system to identifya hand-held baton and determine itsorientation, so audience members couldvote by holding their baton eithervertically or horizontally. The batonswere covered in a highly retro-reflectivematerial, illuminated by a ring of LEDsaround the camera lens, allowing thesystem to operate independently ofthe ambient light level.

Surround videoWe have been looking at ways to useimaging technologies to present newforms of content to our audiences.Most people are already familiar withthe concept of surround sound, whereadditional speakers are deployed aroundthe room, fed by additional soundchannels. We have been investigatingwhether a similar approach could beapplied to video.

Rather than simply looking at the useof a very large screen display, we haveinvestigated the concept of augmentingan existing conventional display with animage projected onto the walls andceiling of the room. The projected imageis generated in such a way as to extendthe field-of-view of the existing display,helping to provide a ‘context’ for themain image, such as additional motioncues. We recorded a range of sequenceswith two rigidly-coupled cameras, onefitted with a conventional lens and theother with a wide-angle fish-eye lens.The image from the first camera was

shown on a flat-panel display, whilst theimage from the second was projectedonto the walls and ceiling, using aprojector and a spherical mirror. Real-time image processing was used tocompensate for the distortion introducedby the fish-eye lens and projectionsystem, ensuring that it aligned in positionand scale with the image on the flat paneldisplay. We also investigated severalmethods of synthesising a surround imagefrom existing content, shot without asecondary camera.

Initial tests have been very promising,with audiences reporting a much greatersense of immersion.

3D productionplanning softwareThe quality and ease of use of 3Dcomputer modelling has now reachedthe point where we can use it for pre-production planning. We are workingwith some of the developers of thissoftware to see how it can best beapplied to making production morecreative and to reduce costs. We havestarted with studio layout and set design,although there are also other potentialapplications such as training.

Producers can test different options ina virtual space, helping them to makeeffective choices earlier in the productionprocess. Our relationships with thesoftware developers have allowed usto model real studio facilities such ascameras and cranes. This enablesproducers to see the exact shots theywill get using particular equipment, sothey can hire in without over-specifying.

Some programme makers have tried outthe set design process. They found it notonly useful in visualising the design, butalso in communicating their ideas to othersin the team. Decisions can be convertedto floor plans for creating and buildingthe real set. The virtual model is alsoeasier to store than the traditional card-board models, and can be retrieved, forre-use or as a reference for the next series.


DD--33 VViiddeeoottaappee Preservation System

The BBC Television and Radio

archives are among the largest

collections of broadcast

content in the world. They

contain more than 300,000

hours of radio and 600,000

hours of television, and these

figures grow daily.

The archives hold content on a varietyof recording formats, which reflecttechnological advances made over theyears. Great importance is placed onbeing able to retrieve content of anyage, and it is often necessary to maintainaccess by transfer of content held onold formats to newer ones, because ofphysical media decay or obsolescenceof replay equipment.Transfers mustbe done in real time with humansupervision, because a sub-optimumreplay or poor conversion will irreparablydamage the content. With such asizeable archive, this can be costly.

In an attempt to improve matters, theadvantages of migrating to file-basedformats have been explored. Whencontent is held as a data file the contentformat is no longer dictated by thestorage medium, and in future it shouldbe easier and faster to move to a newstorage medium, perhaps at the touchof a button. Content on file servers ismore accessible than that held onvideotapes stored on shelves. Qualitydegradation ought no longer to be anissue because true digital clones canbe produced from data files.

Content held on D-3 videotape is nextin line for preservation. The D-3 formatis only 16 years old but is no longersupported by its manufacturer. The BBCholds about 380,000 D-3 tapes (somecopied from the earlier analogue two-inch quadruplex format) and 100,000have been selected for content retention.

We have devised a system to capture thecontent from the D-3 tapes, and toaugment it automatically with metadata.

The system is based around a high-endPC with an SDI capture card andspecially developed software. D-3 isa digital composite PAL format, andBBC-designed Transform PAL decodersare used to give the best possible videoquality. The captured signal is written intoan MXF (material exchange format) file,together with descriptive and technicalmetadata that the system collectsautomatically from the BBC’s ‘INFAX’records – for example, programme andseries names, transmission date andduration. The system also adds the errorlog from the tape replay, and analyses thevideo signal for picture flashes and certainspatial patterns, to note the level ofcompliance against OFCOM recommen-dations for photosensitive epilepsy. It alsoproduces a low bit rate ‘browse quality’copy of the content, and includes soft-ware for performing quality checkingof the MXF file.

The system uses open standards whereverpossible, to avoid technology ‘lock-in’ andto ensure that the files can continue tobe accessed and understood. All thesoftware written by the BBC in this workhas been released as open source.Thisis to promote interoperability and toencourage further development by otherparties, in particular to exploit the featuresof MXF for searching the content.

Initially the files are to be stored on LTO-3data tape.When costs permit, they maybe moved to file-servers. Up to tenprogrammes can be stored on each tape,a forty-fold saving on physical space overD-3, even though the video data is helduncompressed. LTO-3 is by far the mostpopular data tape format on the market,and is supported by a number of manu-facturers. It has an open architecture, andUltrium, the consortium responsible forthe LTO-3 standard, has decreed thatnew generations of the data tape drivesmust be able to read tapes from at leastthe previous two generations.

‘‘ TThhee ssyysstteemm uusseess ooppeenn ssttaannddaarrddsswwhheerreevveerr ppoossssiibbllee,, ttoo aavvooiidd tteecchhnnoollooggyy‘‘lloocckk--iinn’’ aanndd ttoo eennssuurree tthhaatt tthhee ffiilleess ccaannccoonnttiinnuuee ttoo bbee aacccceesssseedd aanndd uunnddeerrssttoooodd..’’

PR

OD

UC

TIO

N


Planning for music showsOur work in previous years hasproduced a production planning systemaimed at making the preparation of thecamera script for a music show moreefficient. It uses beat tracking to link themusic breakdown to the lyrics and thento a standard video editing package, toallow camera shots to be planned againstthe music. Once the planning is completea camera script is produced automatically.

We have now refined the software anddeveloped quicker ways of working withit.We have encouraged more widespreadtesting of this improved version byproduction assistants and directors tosee if the benefits can be achieved inpractice.This has brought some mixedresponses. However, interest is growingand after a little guidance and ‘hands-on’experience, some directors are nowtrying the system with a view to usingit for their productions.

In working more closely with productionteams, we have also come to understandbetter what they require for differenttypes of music show.This has led us todevelop a more straightforward systemwhere only a relatively simple annotationneeds to be added.This is typical of

shows that have a large number ofmusical items, where full planning of eachwould be too time-consuming. As wellas beat and bar indications, our simplersystem allows markers with commentsto be added anywhere in the musicbreakdown, and the standard productionscript is later generated from this.

We now have music production teamstesting this simpler system, and we hopeto be able to work with them on somemajor music productions over theforthcoming year.

Tapeless recorderWe have made several enhancements tothe tapeless recording system that wetrialled last year with the BBC Children’sprogramme BAMZOOKi.

We have now produced a tapelessrecorder capable of recording four videoand 16 audio channels simultaneously,compressing them in real-time to therequired production format (typicallyDVCPro50 or Avid ‘2 to 1’), and storingthem as MXF files. Although a PCequipped with dual-processors, and eachwith dual-cores, is required to achievethis, the overall cost is on a par with amid-range server and is dominated by

four SDI capture cards. By equipping therecorder with 2 Tbytes of hard disc, itcan record for up to two or three daysin a studio.

Low cost serverA further development of this work hasbeen a video server based on the widely-used SAMBA file system.This offers a wayin which a low-cost server can be usedto share content across a cluster of editsuites that are using either Avid or Appleediting software.

There are two difficulties in sharingcontent in this way, which we have beenable to overcome by exploiting thevirtual file system extension capabilitieswithin SAMBA. Firstly, each Avid editingclient expects exclusive use of theexternal storage. Shared access bymultiple Avid clients leads to conflict,as each one attempts to create its owndatabase files to index the video contentthat it finds. Our configuration of theserver’s virtual file system keeps thesefiles separate, so no conflicts arise.

The second obstacle is that not allediting systems can work directly withMXF files – for example, Apple’s FinalCut Pro.To overcome this, we have

implemented on-the-fly unwrapping ofthe MXF files within the SAMBA server.This again exploits the virtual file systemextensions of SAMBA, but in this case,when the original MXF files are accessedby the editing system, the serverremoves the outer ‘MXF’ layers andprovides only the encoded content,e.g. DVCPro50, from within. In this way,the MXF files held on the server appearto the editing system as DV encodedfiles, which it can then use directly.

The next stage in the production flow isto make the recorded content availablein a convenient way within the editingsystem.To achieve this, we are makinguse of the AAF (‘Advanced AuthoringFormat’) standard.When required, anAAF file is produced which contains thedetails of all the MXF files recorded inthe studio over the interval of interest.Opening the AAF file in the editingsystem then populates the editing ‘bin’with this content, making it immediatelyavailable for editing.

Additionally, for multi-camera productions(where each camera view is recordedsimultaneously), the resulting recordingsare automatically grouped together, whichsaves time by avoiding the manual step of

grouping and synchronising therecordings within the editing system itself.

A typical multi-camera productionmethod is to produce a ‘Main’ recording,which includes cuts between the camerasperformed by the vision mixer, andseparate, isolated recordings of somecameras (‘Isos’).These latter recordingsmay be used in the editing of theprogramme to give different views,e.g. an audience reaction shot, whenappropriate. However, a drawback of thismethod is that the timing of the cutsdone as the recording is underway maynot be ideal and slight trimming in post-production may be desirable.We canoffer this capability by recording thetimings of the cuts applied by the visionmixer, and from these, creatingcorresponding edit entries in the AAFfile.When this AAF file is opened, theediting timeline will show the originalcamera recordings with the editsbetween them, which can then betrimmed where necessary.

To understand better how thesedevelopments perform in practice,we are working alongside the productionteam for the BBC programme Eastenders,recording with our tapeless system in

parallel with their normal tape-basedprocess. It is too early to have clearresults but the advantages of easy accessto the recordings are obvious, withimmediate replay into the control roomand studio being a notable benefit. In theedit suites, rapid access to the recordingsalso saves significant time. In this case,the use of the MXF standard forrecordings is a particular advantage asthe files can be used directly within theediting system, eliminating the need fora file import stage. Although not as timeconsuming as ingesting from tape, fileimport would still slow down the processand reduce efficiency.

Editing from P2memory cardsAnother development was at therequest of BBC Sport. For the coverageof World Cup 2006 they needed to beable to edit content from Panasonic P2camera memory cards in an Avid editor,and transfer the edited version back toa P2 card for playout on location.Thislast step was proving impossible becauseof the way MXF is used on P2 cards.After studying the problem, we wereable to produce an application toperform the necessary conversion andcopy back to the card.The application

Programme Production

Our emphasis has continued to be on exploiting commodity IT

hardware so that the cost of equipment used for production can

be brought down.

Where, in the past, four videotape recorders might have been used

in a studio, the trend of increasing computing power means that a

single mid-range PC can now perform this entire task.

This capability will bring about a dramatic change to the way pro-

grammes are recorded and pass through the production chain. Hence,

we are continuing to work closely with colleagues in programme

production, so that we can understand their requirements and see

how the potential of the new technology can best be exploited.PR

OD

UC

TIO

N

Programme Production continued

was used with great success, and meansour production colleagues can now workmore flexibly with P2 cameras.

We have made our software available asopen source so that others can benefitand share in its development.Thesepackages, along with related software,can be found at the ‘Ingex’ project pagehttp://ingex.sourceforge.net/

This work is described further in a paperthat was presented at IBC 2006, wherewe also demonstrated the completesystem in the New Technology Campus.

As our work in this area progresses,we are continuing to explore with otherprogramme production teams how wecan work with them to improve ourmutual understanding of therequirements and benefits of tapelessproduction techniques.

FloormanFloorman is a wireless picture monitordeveloped by BBC R&I which may beused in one of two ways, with differenttechnology for the two applications.We are now seeking to license thetwo versions.

• On the studio floor or on location,by a producer/director, usually inpreparing and rehearsing multiplecamera shots. It is planned tocommission a manufacturerto produce and install a licensedsystem in a BBC studio.This willbe available for use by productionteams, to allow further assessmentof its market potential.

• By a presenter at a live outsidebroadcast, for reviewing the actionon which they will be commenting.Our prototype has been usedregularly by horse racing presenterClare Balding, and BBC OutsideBroadcasts are currently using it fora market trial. If the trial is successfulwe will enter into technicaldiscussions with a manufacturer tohand over licensed designs.

Timed text andsubtitle multicastingThis covers our work in handling subtitlesand similar text information where timingdetails are also included.The work is intwo main areas.

Firstly, we have been providing thesoftware tools to convert broadcastsubtitle files to the formats required foruse by the iPlayer, and for programmesprovided online via bbc.co.uk. Subtitlesused in each of the trials of iPlayer havebeen originated in this way and wehave continued to refine the conversionprocess as the user requirementshave evolved.

Secondly, we have continued toparticipate in the ‘Timed-Text’ workinggroup of the World Wide WebConsortium (‘W3C’), which is in thelate stages of developing a standard forcarrying timing details with text.This isapplicable not only to subtitle exchangebut also production processes,e.g. associating scripted text or indexingdetails with the recording of the content.

Building on this latter work, we areworking within an AAF Associationand EBU subtitling adhoc group whichis producing a standard means forconveying subtitles in AAF and MXF files.This is based on the forthcoming W3Ctimed-text Recommendation but furtherwork to specify details within the AAFand MXF standards has also beenrequired to complete this functionality.

22 APRIL 2006 – MARCH 2007 23

The past year has seen great changesand challenges for wireless production,with the closure of our favoured bandfor PMSE (Programme Making andSpecial Events), 2500 to 2690 MHz.This loss of spectrum together with theincreased popularity of radio cameras,the risk of yet more losses, increasingspectrum costs and the desire to makeprogrammes in HD are presenting realchallenges to the industry. Alternativespectrum between 2 GHz and 3.5 GHzis now being used instead, but thisspectrum is fragmented and congested.Interference from high power CDMAservices is causing concern at many sites.

An additional worry is the spectral purityof higher power COFDM systems thatuse these microwave bands. Commerciallyavailable systems have been measured,and some found to radiate excessive out-of-band energy in the adjacent channels.This limits adjacent channel operation,effectively halving the number of fullperformance channels that can be usedat any one time.

BBC R&I has been working closely withJFMG Ltd (the company that managesthe spectrum on OFCOM’s behalf forprogramme making and entertainmentuses) to identify these problemsand alert the relevant manufacturers.

A measurement programme also lookedat protection ratio issues and the effectsof 3G CDMA cellular services on currentreceivers. By working with industry inthis way, we hope to improve theperformance of commercial equipment

allowing broadcasters to make betteruse of the limited PMSE spectrum thatremains available.

Growth in broadcast applications, andthe competing demands for 2 GHzspectrum from mobile and WiMaxapplications, is prompting us to considera move to higher microwave bands.Initial work at 10 GHz has identifiedimplementation, link budget and Dopplerissues.We are now focussing on the7 GHz band, working in collaborationwith UK manufacturers on a solution fornext generation HD Wireless Production.It is planned to exploit MIMO researchwork, described later, to provide newtechnology for PMSE use in thesenew microwave bands.

Radio Spectrum for Production


We are looking at ways for

the BBC to continue to

improve and expand its use

of digital radio cameras despite

changes in the spectrum we

license from OFCOM.

‘‘ WWee aarree nnooww ffooccuussssiinngg oonn tthhee 77 GGHHzz bbaanndd,, wwoorrkkiinngg iinnccoollllaabboorraattiioonn wwiitthh UUKK mmaannuuffaaccttuurreerrss oonn aa ssoolluuttiioonn ffoorr nneexxttggeenneerraattiioonn HHDD WWiirreelleessss PPrroodduuccttiioonn..’’

PR

OD

UC

TIO

N


To these ends a significant re-engineeringof our digital television delivery systemhas been started which will result ingreater multiplex efficiency and networkresilience. It will also greatly simplify boththe procedures involved in Switchoverand the BBC’s general ability to maketechnical changes to its digital services.We have contributed to the technicaldesign of the new system, working withour technology partner Siemens andcolleagues in BBC Operations.We aremaking extensive use of our digital TVtest system to evaluate suitabletechniques and technologies, modellingvarious multiplex occupancies to assesspicture quality and strategic flexibility(the ability to change the compositionof the multiplex for new services orservice components, or to enhanceor transfer existing ones).

Helping the viewersAs the analogue transmitters areswitched off, many digital transmitterpowers will be increased and thedigital coverage is predicted to besimilar to the current analoguecoverage, approaching 100%. However,it is estimated that about 10% ofdomestic aerial installations will needattention to receive the digital

transmissions. It is therefore importantto give the best possible guidance toviewers to help them decide if theiraerial needs to be checked.

The reception of existing analoguebroadcasts can be a very useful guideto the condition and effectiveness of adomestic aerial installation, and we havebeen researching several techniquesbased on this.We first tried a test patternthat was designed to become lessapparent as signal quality fell.Whilst thistechnique showed the correct averagetrend – those with good aerials tendedto correctly identify the pattern – therewas too much spread on individualresults to give a reliable indication.

We have now created a test patternusing teletext. This gives an objectiveresult, because the pattern is eitherreceived correctly or with obviouserrors. The pattern has been designedto be as resistant as possible to delayedimage interference (otherwise known as‘multipath’ or ‘ghosting’), which normallyaffects teletext reception significantly, butwhich the digital television transmissionsare designed to ignore. It also appears tobe reasonably tolerant of co-channelanalogue signals.

Tests have been successsful, with errorsstarting to show just at the point wherean aerial installation ought to be checked.The test signal is about to be transmittednationwide on the four original analogueservices. (It would not be appropriate totransmit it on the analogue ‘five’ service,as the transmitter powers are at lowerlevels than for the other services, andthe results would be misleading).

Our work in this area forms a partof the overall work being conductedwithin the Digital Reception PredictionGroup run by DigitalUK, the non-profitorganisation leading the process ofdigital TV switchover in the UK.

The digital switchoverassisted help schemeFor the elderly and those withdisabilities, Digital Switchover will presentsignificant challenges. In September 2006the UK Government announced theDigital Switchover Assisted HelpScheme, also known as the TargetedHelp Scheme.This will offer people over75 and those with significant disabilitieshelp in the form of new receivingequipment, installation and follow-upsupport. Central to the scheme will bethe procurement of equipment designed

particularly for these vulnerable viewers,to promote accessibility to digitalservices for everyone.

Building on specific needs identifiedby the Consumer Experts Group(representing the elderly and thosewith disabilities) and in consultation withthe DCMS, DTI, Digital UK and receiverequipment manufacturers, we havecollated and produced a set of corerequirements.These are intended tomaximise ease-of-use for the greatmajority of those eligible for help,without unduly prescribing any particularimplementation that might be offered ina process of open tendering, and withoutstifling ingenuity and innovation.

After a period of consultation the DCMShas published an invitation to tender forthe supply of equipment on the basis ofthese requirements. (The 2006 CoreReceiver documents are available onthe UK government’s digital switchoverwebsite: www.digitaltelevision.gov.uk).

Following the Licence Fee settlement, theBBC and DCMS are currently negotiatingthe commercial details of administeringthe Assisted Help Scheme, whilst someof the technical details of satisfying the

Core Requirements are under discussionin various industry fora.

Impulsive interferenceExperience has shown us that the effectsof impulsive interference need to beallowed for when specifying or designingsystems for digital terrestrial television.Impulsive interference can come froma wide variety of sources, for examplecar ignition systems, electric motors,light switches or room thermostats.Typically, it breaks the picture up intoa mosaic of blocks, or causes amomentary loss of sound.

We have developed a receiver thatmeasures the effects of impulsiveinterference against two key trans-mission parameters, signal level andtransmission mode.The receiver canbe left logging its results on a laptopPC over many days, to allow us tobuild up a statistical model.

The assembly of a pair of these receivershas only recently been completed, andthe next stage will be to deploy themin domestic environments to makemeasurements under a rangeof conditions.Tests in the laboratoryhave already confirmed one expectation

concerning transmission modes. Thereis a significant benefit in terms ofresistance to impulsive interference byswitching from the 2000 carrier mode(‘2K’) to the 8000 carrier mode (‘8K’),which we hope will be one of thechanges implemented during switchover.

Mobile and handheldtelevision broadcastingWe have maintained a watching briefacross the various emerging technologiesthat could be used for mobile andhandheld television broadcasting, andprovided advice and guidance whereappropriate. If mobile transmissions areimplemented using a cellular networkradiating in the spectrum released byswitching off analogue television, thereis the possibility for interference toexisting digital television reception.We have made measurements toanalyse the conditions under which suchinterference will occur. We have kept therelevant standards organisations (mainlythe DTG and EICTA) informed of ourresults and this should allow the effectsfrom this type of interference tobe minimised.

DE

LIV

ERY

Digital TV – Switchover

From autumn 2007, UK broadcasters including the BBC will start

to switch off their analogue terrestrial television signals.This

process, known as ‘Digital Switchover’, will take place progressively

across the UK, and is expected to be complete by 2012.

During this process the two digital terrestrial television (DTT)

multiplexes the BBC operates are expected to migrate from

16-QAM to 64-QAM modulation, with a consequent increase in

deliverable bit rate, and our services will inevitably continue to

develop and evolve.We have also been asked to accommodate

services from several other public service broadcasters in

our multiplexes.

DE

LIV

ERY

http://www.digitaltelevision.gov.uk/pdf_documents/publications/2006/Core_Receiver_Requirements.pdf

Digital TV – Architectures

FreesatThe BBC has been investigating thepotential for a free satellite service,to complement Freeview in areas whereDTT is currently unavailable. From ourexperience of launching Freeview weknow that most of the technical buildingblocks for a free satellite service alreadyexist. However there are significantdifferences between DTT and DSATthat we are having to take into account.Another important part of our work hasbeen to define appropriate behaviour fordifferent classes of receiver (for example,some receivers may have storage or becapable of decoding HD transmissions)and to consider the test requirementsfor each of these. Currently the decisionon the future of Freesat is under reviewby the BBC Trust.

Centralisation of codingand multiplexingThe BBC is planning to centralise theequipment that codes and multiplexesits digital TV services. In advance ofthis, we have taken a significant rolein the testing, configuration andtechnical sign-off for coding andmultiplex equipment in BBC Scotland’snew headquarters in Pacific Quay.This experience is proving valuableeven though we know the technologywill advance in the meantime. Duringthe coming year, working with ourtechnology partners Siemens, wewill be setting up a new test-bed atKingswood Warren to explore theoptions for the architecture andequipment for the centralised system.

DTT Multiplex Bre-engineeringIn summer 2006 the BBC decided thatBBC Parliament should be played outon DTT in full screen, instead of thequarter-screen ‘red button’ service thathad been operating since the launch ofFreeview. The only way to do thiswithout sacrificing the picture qualityof all the services in our Multiplex B wasto change the coding and multiplexingsystem for newer and more efficientequipment. This meant going to anothersupplier for all the major technicalfunctions, including the primary encodingof the video and audio components, themultiplexing, and the capacity switching.Capacity switching was being used tocarry CBeebies during the day andBBC Four at night, using the samecapacity. With the new equipment itproved possible to extend these time-sharing arrangements to other services,in particular to make some capacityavailable for the trial of BBCi catch-up.The improved efficiency of the newequipment has also allowed theBBC to increase the number of NewsLoops from two to four. We modelleda number of different options to helpour colleagues in BBC Distribution decideon the final capacity allocations, beforehelping to plan the changeover and putit into effect.

26 APRIL 2006 – MARCH 2007 27

The ARENA projectThe BBC has joined with nine othermedia organisations in Europe in aEuropean Union collaborative project,ARENA, to try to devise a system thatproduces audience figures consistentlyacross all these services and platforms.[Note that viewing data is collected onlyfrom household members and individualswho have their given informed consentto allow this.]

ARENA is planning four field trials ondifferent platforms, linked through acommon audience data collectionsystem. The BBC has a particular interestin one of these field trials, on the newBT-Vision platform where we supplysome of the on-demand content, aswell as operating two of the DTTmultiplexes also receivable throughthe BT-Vision ‘V-box’.

This is an ambitious project, and theprocesses will be simulated beforethe system itself is built. Our work sofar has been to establish the businessand functional requirements for cross-platform audience research, talking toother organisations too to try to predicthow services will develop. The projecthas used this information to create aformal data model. Our work is nowto ensure that the model is a goodrepresentation of the real-worldsituations and that it is tested correctly –the eventual outcome from the otherpartners will be software that can beported into set-top boxes or theplatform operators’ distribution systems.

Audiences forinteractive servicesAt the request of BBC MC&A, and inparallel with the formation of ARENA,we have been developing an idea tomeasure ‘red button’ interactive TVusage on our digital television services.This uses a small ‘index mark’ added toeach page of interactive content, createdby the video rendering engine in the

digital set-top box or digital TV usinginstructions conveyed as part of theinteractive page description sent tothe receiver. Last year, we produced aspecification for the index mark on thebasis of laboratory tests, and theBroadcasters Audience Research Board(BARB) commissioned a prototypedetector from their technical contractorAGB-Nielsen. About 200 of thesedetectors have now been installed inthe homes of existing BARB panelmembers (these are households thathave agreed to have their viewingpatterns monitored by BARB), for alimited trial. With the co-operation ofother UK broadcasters, the index markswere added to some interactive contenton terrestrial and satellite broadcasts,and also on to a digital cable system.Tests in collaboration with AGB-Nielsenwere successful, after some minoradjustments to accommodate variationsin the different signal chains.

The index mark is designed to be justoff-screen on the majority of televisionsets, and indeed, no complaints aboutvisibility were received, althoughpotentially all digital receivers willreproduce it, not just those in the BARBtrial. BARB is currently re-tendering itstechnical contracts and a decision onthe use of this technique will dependon the outcome of this process.

Audio watermarkingAudio watermarking, i.e. hiding infor-mation in the programme sound,has been tried unsuccessfully for radioaudience measurement in previousyears. The systems offered have beenfound to produce unacceptableimpairments to the sound quality. Finally,this year, a system that had been testedseveral times before was found to havebeen improved sufficiently to be inaudiblein our laboratory assessments. On-airtrials were able to start with much lessworry about the possible negative impactthan has been the case previously.

Audience Research


How do you measure audience

viewing figures when your

audience has many ways of

watching your programmes?

This is the challenge facing

BBC Marketing Communications

& Audiences, with broadband

and mobile phone access

becoming increasingly popular

and time-shifted viewing

getting easier. Programmes

are augmented with subtitles,

interactive content, and elec-

tronic programme guides, and

we need to track usage of

these too, if we are to under-

stand our audiences properly.Designing for changeWhen Nicam stereo sound and teletextwere added to analogue televisionbroadcasts, this could not be donewithin the existing signal standards.Although the changes were madesuccessfully, considerable engineeringingenuity was needed and there wereconstraints on what could be achieved.Viewers had to buy new TVs if theywanted the new features, and the extrasignals had to be added in a way that didnot affect reception on existing sets.

When the DVB standard (the inter-national standard for digital televisionbroadcasting) was drawn up, theengineers responsible made sure itcould be extended in a systematic wayto incorporate new features.Thestandard also defined a ‘multiplex’structure that makes it possible to add,modify and rearrange TV channels andother broadcast components.This meansthat for many of the changes viewers donot have to replace their receivingequipment. Changes are picked up by

digital TV receivers automatically, orby ‘rescanning’ or downloading newinternal software.

The BBC continually exploits thisflexibility to develop its digital services.Since the original launch, BBC R&I havehelped to develop a number of newfeatures, including interactive services,enhanced electronic programme guides,new regional broadcasts, and audiodescriptions of programmes for visually-impaired viewers.

However, a great deal of planning isneeded even for a small change,especially as we always try to avoid anyloss of service. BBC R&I continues tocarry out some of the developmentwork, and to give advice and practicalassistance. Using our experience and ourspecialised test facilities, we make surenew features work as intended, any newequipment is suitably chosen, and thebest use continues to be made of thedigital multiplex capacity.

DE

LIV

ERY

Digital Radio MondialeDigital Radio Mondiale (DRM) offers clear, high-quality audio and data services onthe long, medium and short wave bands. DRM is now a fairly mature standard andthere are many hundreds of hours of transmissions on-air.

Consumer receivers are crucial to the large-scale uptake of DRM, and are beginningto become available.We have driven this development in two key ways.

Firstly, all the current receivers are based on code which we originally developedand licensed to manufacturers, and we continue to support our licensees withtheir implementations.

Secondly, we have built up an extensive receiver testing facility at Kingswood Warren,allowing us to perform tests including sensitivity, selectivity and linearity. In particularwe have been able to expose receivers to known field strengths in order todetermine if they will work in the field with the signals that broadcasters plan todeliver. This has revealed a number of shortcomings in the early prototypes, andwe have worked closely with manufacturers to improve their performance.

We have also continued our development of receiver algorithms, looking particularlyinto the possibility of seamless AFS (alternative frequency switching).This couldenable a low-cost receiver with only one front-end to take advantage of the benefitsof multiple-frequency transmissions, which we have previously demonstrated withour dual-front-end diversity receiver.

Whilst the BBC World Service is our main customer for DRM, we have recentlyhelped BBC Distribution to prepare for a consumer trial of DRM on medium-wave,being carried out in partnership with BBC Radio Devon and National Grid Wireless.We have given advice based on our long experience of DRM as well as providingmonitoring stations for the trial. As part of this we have developed a monitoringreceiver network to allow reception conditions in various locations to be observedin real time.

28 APRIL 2006 – MARCH 2007 29

We started to develop Kamaelia abouttwo years ago, and at the end of 2005BBC Radio & Music Interactive had usedit to produce a prototype system fortranscoding all BBC radio output.

Over the past year, we have created atelevision equivalent for BBC New Media.This was a significant piece of workneeding a large number of optimisationsto produce a scalable system. Since thenKamaelia has been used in a number ofother projects.Tools for re-framing videofor mobile devices were created forresearchers looking into mobile delivery.A system was created for capturing BBCbroadcast content, storing it, transcodingfor different bit rates, and preparing it forpeer-to-peer transfers. A rather differentapplication, cross-site white-boarding toallow people in different locations tocollaborate more easily, was quicklycreated, then expanded to include audioand made peer-to-peer aware with localremixing for scalability.

We continue to move Kamaelia out ofthe laboratory and into the wider BBC,academia and industry.We have giventalks at a number of high profileconferences for software development,

and we ran a workshop in Brussels forinterested collaborators.We mentoreda number of students during Google’sSummer of Code 2006, resulting in toolsfor peer-to-peer online distribution, 3Duser interfaces, and proofs of conceptfor trusted communications.This alsoproved that Kamaelia can be learnt andused effectively in a very short periodof time by the average developer.

Our next steps are to consolidateKamaelia’s core and to increasetestability, looking for a similar levelof verifiability to hardware.We are alsoaiming to make some aspects ofKamaelia to the non-programmer –we want to take the graphicalconstruction tools that are used inKamaelia to sketch new systems, andextend them to create the code directly.

Because Kamaelia is open source, thisenables everyone from schools andcolleges through to enterprising membersof the public, to use it to createapplications including ones to repurposeour content to serve their needs.

To collaborate with us, visithttp://kamaelia.sourceforge.net/Home

KamaeliaDAB and Digital Radio Mondiale


AAC coding and DAB+BBC R&I submitted one of severalproposals in response to the WorldDMB*call for contributions for adding a moreefficient audio coding system to DAB.The work was co-ordinated by theWorldDMB Technical Committee in acollaboration across Europe, Asia, NorthAmerica and Australasia.

Our proposal was adopted byWorldDMB after a careful analysis of allthe competing solutions and after fieldtests carried out in the UK and Australia.

The result is DAB+, announced in March2007.We have taken responsibility forstandardising DAB+ through ETSI.

The main part of the proposal is anoption to use high-efficiency AAC codingfor radio broadcasts, creating capacitythat could be used for more stations,interactive services or surround-soundbroadcasting. Audio quality is improvedat lower bit rates, and an improved errorcorrection scheme gives slightly greatercoverage, quicker tuning and no relativetiming delays between stations.

For the BBC this work is decidedly longterm, to ensure that technical standardsthat may one day allow us to introducenew services are carefully consideredbefore being agreed.While countriesthat have yet to launch DAB could optimmediately for DAB+, the BBC remainscommitted to its MPEG Audio Layer IIbroadcasts. It is well aware of itsresponsibility to its listeners, many ofwhom will only recently have boughtnew DAB radios.

Kamaelia is the software

toolkit we created for

exploring on-line delivery

methods for the BBC’s

archives. How do you make

the BBC's considerable

archives available on-line to,

say, 20 million households at

once, especially if they all want

to watch something different?

The solution needs to be

scalable, acceptable to

organisations that have rights

over content the BBC holds,

and flexible enough to exploit

changes in technology.

Kamaelia, parallelism and concurrencyDistributing the BBC Archive content online is an inherently parallel process. Largenumbers of people are simultaneously searching and downloading material, and this isa huge test of a distribution system, both its hardware and software.

Software engineers define concurrency as the real or apparent running of multipleprocesses at the same time. Modern computers, particularly professional machinesand games consoles, have multiple processor cores, whereas computer software hasits roots in single processors and sequential processes.The reason that concurrencyhas been made to work is down to putting fundamental rules in place.

Kamaelia was developed by BBC R&I to make an online distribution systemstraightforward to create and easy to maintain.A happy side effect is that it can beused to develop other applications for communication where the biggest problemsarise from parallel processes. It is this aspect that we are now exploiting, andencouraging others to exploit by making Kamaelia open source.

DE

LIV

ERY

* Formerly the WorldDAB Forum.

http://kamaelia.sourceforge.net/Home


WiFi for TVprogramme productionWiFi is rapidly becoming a de factostandard for connecting Media Centrecomputers to TVs and audio systems.Thisis not an application WiFi was originallydesigned for, but some additions to thestandard have made it just about suitablefor domestic use.We have been lookingwith a number of companies at otherways of extending the standard, to allowus to use WiFi in TV production.

WiMaxWiMax is similar to WiFi, but works athigher powers and has greater range andcapacity. Its transmission in the 5.8 GHzband is licence-exempt, and we havecarried out a series of tests to see if itcould be used for ad-hoc contributionand distribution links.Transmissions at thisfrequency can be affected by buildings,so WiMax uses COFDM to benefit fromreflected signals.

We arranged for a transmitter aerial tobe installed on a mast on high ground inNorth London, and mounted a receiveraerial on a 10 m mast on a radio van.Tests at points chosen at random within5 km of the transmitter all gave athroughput of at least 10 Mbit/s, and

some achieved 20 Mbit/s.These bit ratesare all adequate for high-qualitycontribution and distribution circuits.

We then moved to the Westminsterarea, where BBC News are trying tofind a quick and economical way ofsetting up links back to their local studio.UrbanWiMax, a company that providesInternet services for business users,provided us with a temporarybidirectional link.Their WiMax networkhas a nominal maximum capacity of4 Mbit/s (limited by the backbone ratherthan the WiMax itself), and is intendedto provide spot capacity and diversityfrom cable connections. It covers mostof the locations in Westminster thatBBC News regularly use.

This is a much more built-up area,and tests with the radio van showedsome locations where no signal wasreceived. However, where a signal wasobtained, in most instances it wouldsupport up to 2.7 Mbit/s video and192 kbit/s audio streamed simultaneously,definitely acceptable for news reportswhere a lower picture quality isacceptable and a video bit rate of about1 Mbit/s is usually considered adequate.We also noted that the audio quality

exceeded our expectations even whenreduced to 64 kbit/s.

Audio over IP In 2004 some of the BBC’s local radiostations started to exchange audio contentover standard IP networks, using broad-band ADSL connections.We were askedto help when it was found that somecodecs were more affected by networkconditions than others, and that codecsfrom different manufacturers did notalways work together. This was worrying– we were hoping that IP technologywould be an easy replacement for ISDNlines when these start to disappear.

We have therefore carried out a seriesof investigations, running tests on arange of commercial codecs.We are nowworking to establish an interoperabilitystandard and a recommendation of goodoperational practices.This work is beingdone through the EBU, under whoseauthority the standard and recom-mendations will be published, with helpfrom IRT in Germany and Sveriges Radio.(Sweden is already faced with losing allISDN services in the next two years).It is intended to produce a referencecodec with which to test commercialproducts for compliance.

30 APRIL 2006 – MARCH 2007

IT networks are usually packet switched,and their properties are significantlydifferent from the circuit switchednetworks we have traditionally used tocarry our signals. Being primarily designedfor regular IT traffic, they do not alwaysperform as well when faced with largemedia-content files. Over recent years wehave bought or developed a number ofhighly specialised test tools, and we arecontinually studying the latest offerings tojudge if the claims made are justified asfar as media production is concerned.One of our developments is a softwareapplication that can be ported out tomultiple PCs to simulate a busyproduction office or even an entire site.We are now investigating next-generationmulticast streaming protocols, and, withHD production in mind, the performanceof 10 Gbit/s IP routers.

Through this work, we ensure the BBCunderstands the issues surrounding net-work provision, and specifies appropriatelyfrom our technology supply partners.Weregularly feed back the results of ourtests to manufacturers, or do tests withthem, to help tune products for mediause.We have also found it important totake part in the work of the relevantStandardisation bodies, because of the

strong influence that standards have onthe design of IT equipment.

We also run a number of research andinnovation projects:

PRISMWe lead the DTI collaborative projectPRISM, which is applying distributed-computing ‘GRID’ techniques to mediaproduction. Although the project itselfwill finish at the end of 2008, it is partof our longer term work to try tostreamline and enhance the technicalprocesses in programme production.The other aspect of the work is toencourage standardised interfaces, so thatproduction tools will work together evenif chosen from different suppliers. PRISMwill publish the interfaces it develops asa set of open specifications.

As a simple example of what PRISM isseeking to achieve, imagine creating adocumentary programme from contentwhich has been shot over a period ofyears on different systems. It would helpif the editing system applied anynecessary format conversionautomatically, without the operatorneeding to intervene to take decisions orstart the process.With GRID, the editing

system could delegate the tasks ofcomparing formats and making theconversions, especially if the content wasbeing retrieved from remote locations.

A more challenging example is adding aformal description of media content asit goes into repositories. Descriptivemetadata is important for searching andretrieval. It is however extremely labour-intensive to create, and this is a definitedisincentive. One of the partners inPRISM has considerable expertise onmachine-interpretation of images, andwe are very interested to find out iftheir techniques can be applied here.The interpretation is not entirely auto-matic, but needs only a small amountof prompting on context from a humanoperator. It is nevertheless computationallyintense, and can benefit from GRIDtechniques. PRISM aims to demonstratea scalable high performance system builton a standard IT core.

PRISM is working closely with BBCNorthern Ireland, and will be testedin conjunction with a number ofprogrammes that they produce.

NE

W S

ER

VIC

ES

Modern programme production tools no longer consist of

specialised broadcast hardware interconnected by coaxial cables.

They are now almost invariably computer applications running on

PCs, and they use file-based media storage rather than tape. As a

result, IT network connections are now an essential part of any

production system.

Media files can be extremely large, putting patterns of demand

on an IT network quite unlike conventional IT traffic. BBC R&I

has for a number of years investigated the problems this brings,

and worked with suppliers, manufacturers, and standards bodies

to establish solutions.

Networks for Programme Production

CO

RE

TE

CH

NO

LO

GIE

S


MIMOMIMO is a method for increasing thecapacity of a radio link by running multipletransmissions on the same frequency.To investigate its potential BBC R&I hasproduced a prototype MIMO systemusing two transmitters and two receivers,with the transmissions based on DVB-T.An experimental pair of 50 W ERPtransmitters has been installed at a sitenear Guildford. A survey vehicle has beenequipped to investigate the coverage forboth fixed and mobile reception.Thetransmitters can be switched remotelybetween dual-polarised or co-polarisedtransmission, to explore differentconfigurations.We are doing this work incollaboration with OFCOM, Arqiva andNational Grid Wireless within a new groupset up for the purpose and chaired by theBBC, known as the Advanced TerrestrialTransmission Study Group, or ATTSG.

Initial results using 64QAM rate 2/3 modu-lation suggest that the coverage of dual-polar DVB-T MIMO is very nearly as goodas that obtained for DVB-T itself, with adata throughput of 48 Mbit/s comparedto 24 Mbit/s for the standard system.

Because MIMO is one option that isbeing considered for updating DVB-T,

the results of this investigation are beingmade available to the DVB-T2 technicalworking group (described opposite).

On-channel repeater BBC R&I has developed an on-channelrepeater for the DAB and DVB-Tfrequency bands.This device, first shownby the BBC at IBC in 2005, allows a relaystation to re-transmit a signal on thesame frequency as it is received.This isnormally difficult because the receivingantenna almost inevitably picks upsome of the signal from the transmittingantenna, causing howl-round. Theon-channel repeater is particularly useful

for filling ‘holes’ in the service area ofa transmitter, or extending the range.

We have protected the key technologywith a number of patents, and we haveissued a licence to use the algorithm toa major transmitter manufacturer. Moreenquiries for licences have been receivedboth from the UK and overseas and arecurrently being processed.

Trials have been carried out incollaboration with network providersin the DAB band, with additional DVB-Ttrials scheduled for the near future.Information regarding the behaviour of

the system in complex single-frequencynetworks (SFNs) has allowed refinementsto be made to the algorithms.

This year has seen the original systemenhanced in a number of ways to improvesignal-to-noise ratio and the ability totrack rapidly time-varying coupling paths.So much so, in fact, that the system hasbeen lab tested at 2.5 GHz in a PMSE(Programme Making & Special Events)radio-camera mid-point application.The results are very promising and weare close to having a solution to offer forthis application.This would minimise thenumber of PMSE operations needing twofrequencies at a time when this spectrumis likely to become increasingly expensive.

DVB-T2In 2006 BBC R&I submitted a paper toDVB, proposing the formation of a StudyMission to consider updating thestandard for digital terrestrial televisiontransmissions. The proposal was acceptedand we were asked to lead the study,which identified several technologies thatcould be used to increase both efficiencyand ruggedness.

We were then asked to lead a DVBtechnical study group to continue the

technical analysis, in parallel with anothergroup preparing the commercial require-ments. The work of these groups is stillunderway. Currently it appears that amodest increase in capacity could beachieved through the use of advancederror correction codes, such as thoseused for DVB-S2, together with somechanges to the underlying transmissionsystem to eliminate other minorinefficiencies. Although a significantlygreater increase in capacity (i.e. doubleor more) could probably be achievedthrough the use of MIMO this wouldrequire changes to both the transmitteraerial and the viewers’ receiver aerials.

The commercial attractiveness of theseoptions is still being studied. Existingdigital viewers would need to buy newset-top boxes or digital TVs, so theintroduction of a new standard mighthave to be incorporated into the launchof a new service, for example HDon DTT.

It is expected that a first draft of aspecification for fixed and portablereception will be completed around thebeginning of 2008, with a furtherspecification for mobile receiversfollowing later.

Radio Systems

The BBC is a major user of radio frequency spectrum, not just for

its public broadcasts but also for many internal purposes such as

wireless cameras, wireless microphones, and passing signals back

from outside broadcasts. Spectrum is a scarce resource, and we are

constantly looking for techniques to make more efficient use of it,

as these three pieces of work demonstrate.

CO

RE

TE

CH

NO

LO

GIE

S

In trying to squeeze more capacity out of theprecious UHF spectrum, the DVB-T2 projectis bound by theoretical limits to what can beachieved.The Shannon capacity limit is wellknown; however, it is sometimes forgotten justhow idealised it is. Once certain choices aremade, such as picking a particular combinationof constellation and bit mapping, more detailedapplication of Shannon’s information theoryshows that rather less capacity is then achievable,even in an ideal case.This graph serves as abenchmark for studies of error-correcting codestogether with the possible addition of higher-order constellations like 256-QAM to the system.

MIMOMIMO stands for Multiple-Input-

Multiple-Output and refers to a

wireless system with more than one

transmitter and receiver, all operating

on the same frequency but with each

transmitter carrying a different signal.

Each receiver picks up a mix of the

transmissions, depending on

propagation conditions and the placing

and polarisation both of its aerials, and

those at the transmitters. If the

receivers can determine the mixes in

the form of a mathematical matrix,

their outputs can be passed through

the inverse of this matrix to recover

the original transmissions. In the

experimental system described here,

the receivers calculate the matrix by

measuring the pilot tones already

present in the DVB-T signals.

The increased capacity is obtained at a

cost. Depending on the amount of

mixing, the received signal-to-noise

ratios can be worse than for a single

transmitter and receiver, and this might

reduce range or require more

transmitter power.


Dirac ProWe have now produced an implementationspecifically for higher bit rates, which wehave named Dirac Pro. It shares manyfeatures with Dirac and is most suitableabove 100 Mbit/s.

Dirac Pro’s initial application was to trans-port 50 frames/second (fps) HD signalson the same cables and infrastructureas conventional 25 fps HDTV. Videoat 50 fps conveys a more ‘fluid’ motionand the BBC plans to move to this higherquality standard over the next few years.It is particularly suitable for sportscoverage in HD, and could be usedfor the 2012 Olympic Games. Dirac Proprovides the 2:1 compression, sometimesknown as ‘Mezzanine’ compression,with almost no loss in quality.

Another application of Dirac Pro is thetransmission of HDTV signals betweenstudios and production centres, usingthe existing distribution infrastructure.This requires more compression butDirac Pro is sufficiently flexible to achievethis with little loss in quality. Many otherapplications of Dirac Pro are also possible,in programme production usingnetworked infrastructure (‘desktopproduction’) and also in digital cinema

production. Dirac Pro was successfullydemonstrated to film makers in January2007 at the Hollywood Post Alliance.

Dirac is attracting considerable interestboth from within the BBC and fromother organisations. It was demonstratedpublicly at IBC 2006, where content fromthe BBC’s HDTV trial was comparedfavourably with satellite broadcastsusing H264 at the same bit rate.

We are currently formally standardisingDirac Pro through the SMPTE. A draft ofthe specification, referred to as VC-2, hasbeen submitted and we are now waitingfor it to be ratified. Establishing thestandard will encourage equipmentmanufacturers to incorporate Dirac Prointo their products. An implementationof Dirac Pro, developed jointly with ourcommercial partner NuMedia Technology,will be launched at NAB 2007.

Where next?We now regard our original implemen-tation as a reference and have releasedit to several universities as a develop-ment platform. We are collaborating withthem to exploit Dirac further and pushthe boundaries of video compression.

As 3D displays move closer to reality,both for the home and mobile devices,we are working with the University ofSurrey to develop methods forcompressing 3D video content.Thesetechniques also help in modelling motionin conventional 2D images, and couldhelp in improving compression there.

Brunel University are adapting Diractechnology to produce bitstreams thatcan withstand poor quality transmissionsystems and the effects of networkcongestion, avoiding the completecollapse in quality that bedevilsconventional compression techniques.

As bit rates are pushed ever harder, itbecomes increasingly difficult to maintainpicture quality. Because the relationshipbetween encoder control mechanismsand perceived quality is very poorlyunderstood, encoder designers relyon ad-hoc, indirect, techniques to avoidquality collapsing when encoders arestressed. Manchester MetropolitanUniversity have embarked with us ona project to incorporate psycho-visualmodelling techniques into video encoderdesigns to manage video quality directly,using Dirac software as a test-bed.

34 APRIL 2006 – MARCH 2007

DiracOver the past few years we have drawnon practical experience that dates backto the 1960’s to develop an advancedvideo compression system called ‘Dirac’.Although we originally intended Dirac forhigh resolution images and high bit rates,its wavelet-based compression techniquehas proved suitable for applications fromover 1 Gbit/s down to below 100 kbit/s.It is comparable with the latest standardsH.264/MPEG-4 AVC and VC-1.

Dirac is open source, relies on no otherparties’ intellectual property, and can beused without payment of royalties. Itspotential uses range from low resolutioncoders for mobile phones and Internetdistribution of video clips through toHDTV and beyond with ultra-highresolution Digital Cinema. Dirac worksunder Linux and on Apple Mac PCs aswell as Microsoft Windows.The abilityto work on different operating systemsis important to the BBC’s plans for‘On Demand’ TV via the iPlayer andstreaming content over the Internet.

During the past year we have continuedto improve Dirac.The algorithm has beenmade more efficient and easier to imple-ment. A detailed specification has beenpublished and open-source software madeavailable to allow anyone to developimplementations.A second implementation,called Schrödinger has been optimised forspeed although because of this it is some-what harder to use as a basis for furtherdevelopment. Both implementations candecode standard definition TV in realtime, with Schrödinger able to do thiseven on average specification PCs.

NE

W S

ER

VIC

ES

Almost everywhere video is handled digitally some form of data

compression is applied. Although satisfactory forms of compression

exist for most applications, some are subject to restrictive or

expensive licensing arrangements. This can increase our costs, and

potentially the cost to our audiences in receiving our programmes.

The BBC as a publicly funded body is in a unique position to develop

open compression technology where it has a specific need, and to

make that technology freely available. For example, the BBC’s Open

Archive initiatives will be making content available over the Internet,

and we would like to be able to supply free software to schools and

other educational bodies.

Video Compression – Dirac and Dirac Pro

Measuring video qualityThe technical quality of televisionpictures has been a frequent topicfor discussion, even before the launchof the regular service. The debatecontinues, now largely over digital TVand latterly, HD.

Amongst the early broadcasts of thecurrent HD trial were live programmescovering the FIFA World Cup fromGermany. The broadcast arrangementswere going to be complex using a chainof up to four MPEG-2 contribution linksfollowed by transmission to the homeusing H.264 compression. Knowing that

passing video signals through cascadedcompression systems can cause extrapicture degradation we simulated theentire chain. By experimenting withbit rate settings we were able tomeasure the video quality that wouldbe achievable at home and recommenda number of improvements.

This was perhaps an extreme example,but any production chain will include anumber of different compression systemsbetween camera and broadcast, andthere may also be format conversions.Because of the range of compression

systems now available there is more thanever a need for a reliable and standardisedmethod for measuring video quality.

Measuring video quality is an area weare continuing to research. In the comingyear we plan to establish a test-bed toinvestigate and provide specialist adviceon video quality within the BBC. We arecontinuing our programme of evaluatingcompression codecs and investigatingissues of concatenated coding, focusingmainly on acquisition and production.We are contributing to collaborativework on this topic within the EBU.

CO

RE

TE

CH

NO

LO

GIE

S

Audio bit rate reductionA study has been made of MPEG Audio Layer II coding to find out how good it wouldbe if an optimal coder could be developed. A ‘genetic’ coder was devised that couldexplore the myriad permutations of encoding that could be applied to each 24 msframe of audio.The coder worked more slowly than real time but it was possible torecord the results and listen at normal speed. The work was presented in a paperto the AES European Convention in 2006. We concluded that the potential for furtherimprovement was marginal, the best coders already being close to the performanceof the genetic coder.

There is always pressure to reduce the bit rates of the audio signals on DAB to makeroom for new features. MPEG Audio Layer II coding is adjustable only in moderatelycoarse steps in bit rate, and it is difficult to make reductions without compromisingthe audio quality. BBC R&I has continued to be closely involved in the assessment ofproposed changes.

Multi-channel audio BBC R&I took part in a major series of subjective tests organised by the EuropeanBroadcasting Union’s B/MAE (Multi-channel Audio Evaluation) project group. In the firstset of tests the subjective quality of multi-channel audio bit rate reduction systemssuch as Dolby Digital Plus, High Efficiency AAC MPEG Surround, and DTS was carefullyassessed. These are systems that are, or could soon be, used for broadcast or Internetdistribution of high quality surround sound, with or without accompanying video.

Early results were presented by IRT at the EBU’s Forecast ‘06 conference. A paperhas been written by BBC R&I together with colleagues from the EBU and IRT, and isto be presented at the 122nd AES convention. The conclusions reveal interestingperformance evolution over the ten years since the previous comparable tests. Withnon-critical material, modern coders can achieve results as good as their predecessorsbut at a significantly lower bit rate. However, maintaining consistently high quality stillrequires the same, relatively high, bit rates.

Independent component analysis Independent component analysis is a technique for splitting a sound signal into itscomponent parts, and dealing with each one separately. As a simple example a duetmight be split into the sounds of the two instruments, which in this instance wouldclearly be easiest from a stereo recording. We are fascinated by the applications thatcould be developed from this fundamental technology. A joint project is being set upwith the University of Salford to examine how it might be applied in real broadcastaudio environments.

36 APRIL 2006 – MARCH 2007 37

The Internet revolution encourages media businesses to find new ways to exchangecontent with their customers and between themselves. Digital Rights Management(DRM) is important in controlling and protecting these exchanges. However, someof the systems in use are proprietary, and as the number of connections multiplies,failure to develop interoperable solutions could be a major obstacle to the growthof the industry.

The requirements of DRM vary from limiting access, through ‘copy control’, tomuch more flexible permissions. In some cases DRM is used simply to protect theintegrity of the content or authenticate authorship rather than limit distribution orcollect remuneration.

BBC R&I has been working with partners in industry and academia to develop openDRM solutions. We have supported the broadcast technology providers through theDVB organisation as they specify the technologies for protecting content deliveredover broadcast networks.

We also participate in the Digital Media Project (DMP) as it develops specificationsand builds software prototypes from its ‘Primitive DRM Tools’. These smaller DRMrelated functions can be used to assemble a variety of value chains according toindividual business needs. The DMP aims to provide the technological means andorganisational framework to allow participation in the media industry both by largecontent distributors and individual media creators wishing to distribute content withvarying degrees of security. The DMP specifications bring a more inclusive model forall players in the value chain, from authors, performers, adapters and producersthrough to content providers and home users.

More recently we have begun working with a large group of academics andindustrial partners within a mature and established project, funded partly by theEuropean Union, that has developed a platform for cross media production anddistribution; AXMEDIS.

In working with these international groups to develop and consider the issues ofdeploying DRM solutions we have come to a closer understanding of how ourindustry is changing and can change further to meet the demands of our modernworld. By working with the wider industry we are better placed to control ourown future and meet the needs of a new type of audience in the new digital ageof broadcasting.

Digital Rights Management


‘‘ TThhee DDMMPP aaiimmss ttoo pprroovviiddee tthhee tteecchhnnoollooggiiccaall mmeeaannss aannddoorrggaanniissaattiioonnaall ffrraammeewwoorrkk ttoo aallllooww ppaarrttiicciippaattiioonn iinn tthhee mmeeddiiaaiinndduussttrryy bbyy bbootthh llaarrggee ccoonntteenntt ddiissttrriibbuuttoorrss aanndd iinnddiivviidduuaallmmeeddiiaa ccrreeaattoorrss wwiisshhiinngg ttoo ddiissttrriibbuuttee ccoonntteenntt wwiitthh vvaarryyiinnggddeeggrreeeess ooff sseeccuurriittyy..’’

Audio Compression

CO

RE

TE

CH

NO

LO

GIE

S


Collaborative Projects

BBC R&I works closely with other broadcasters, industry and

universities in order to maximise the effectiveness of our research

effort, to share knowledge, gain early insights, and influence emerging

new systems and standards. Collaboration ranges from funding of

students through to participation in major projects, many part-

funded by government or the European Union.

We are pleased to acknowledge the effort and support given by

all the collaborators and funding bodies mentioned in this review.

The BBC Our research and innovation builds upa depth of knowledge in a technology,usually before it starts being rolled outas an application. When the BBC needsto deliver the technology our expertiseis invaluable to a successful deployment.There are numerous examples in thisReview where we are providing adviceand active assistance, ranging from veryvisible projects such as the DigitalSwitchover and the test transmissionsof HD, to more esoteric work onprogramme production andtransmission technologies.

Other broadcasters The European Broadcast Union is acentral co-ordinating body that bringslike minded organisations together,and many of our contacts with otherEuropean broadcasters are through itstechnical working groups. Extending ourview outside Europe is also important,for new systems are often introducedfirst on other continents, HDTV beingthe best recent example.We havedeveloped stronger links with, forexample NHK in Japan and CCTVin China. These different culturestraditionally view technologies differentlyfrom Europe and we can learn from this.Also these areas, together with India,have a very strong technical capacity andare influential in world markets, and theconsumer marketplace.

Universities Our links with Universities are numerous.As well as direct links, we also have DTIsponsored projects and many informalcontacts.The academic knowledge whichthis brings can underpin any developmentof a new service or system.

Collaboration in Europe We contribute to NEM, the NetworkedElectronic Media initiative, with the aimsof influencing and guiding the researchagenda for the European Union’sFramework programme of research.This is an association of experts in thetechnology of media and their insightand guidance is an important pre-competitive indicator of future directionsthat the industry is taking.

A number of our research projects areenhanced by joining consortia fundedunder the European Union’sFramework 6 research programme.This allows us to contribute and gainfrom projects that are typically budgetedat many millions of Euros.

Arena is part funded by the European Union. It is exploring methods of measuringand aggregating audience figures on mobile phones, broadband connections and videorecorders, as well as the conventional digital platforms. Arena runs for three years andis due to finish in December 2008.We are supported in this project by colleaguesfrom BBC Marketing Communications and Audiences.

AXMEDIS is part funded by the European Union. The project provides an opensolution which builds on technologies and tools to reduce costs and increase efficiencyfor the production, protection, management and distribution of content.

CALLAS is part funded by the European Union.The New Content Paradigms areafocuses on building upon the BBC tradition of rich narrative. We are researching howto bring this BBC quality into the interactive domain, looking at techniques for bothauthoring and play-out of truly interactive (‘open-ended’) stories. CALLAS iscombining the use of emotion-sensing interfaces with interactive narrative techniquesto produce a ‘bardic interactive storytelling engine’.

We lead this DTI-funded project which is developing methods to capture action fromevents such as a football match in 3D and to provide a 360 degree free-viewpointvideo replay of this action, as if from a virtual camera.

MATRIS was part funded by the European Union, and finished in January 2007.It developed markerless camera tracking technology that used naturally-occurringfeatures in the scene. It also developed an inertial sensor and looked at the use of anauxiliary camera to improve the tracking. We coordinated the project demonstrationat IBC 2006, which attracted the interest of several potential licensees.

Participate is a three-year DTI-funded collaborative project. It is exploring theconvergence of pervasive, online and broadcast media to create new kinds of massparticipatory events where a broad cross-section of the public contributes andshares content.

We lead this DTI-funded project which is developing workable production servicesbased on GRID distributed computing techniques.

WO

RK

ING

WIT

H U

S

ARENAwww.ist-arena.org

AXMEDISwww.axmedis.org

CALLASwww.callas-newmedia.eu/

iviewbbc.co.uk/rd/projects/iview

MATRISwww.ist-matris.org

Participatewww.participateonline.co.uk

PRISM

http://www.ist-arena.org/

http://www.axmedis.org/

http://www.callas-newmedia.eu/

http://www.bbc.co.uk/rd/projects/iview/

http://www.ist-matris.org/

http://www.participateonline.co.uk/


Standards

Some standards endure for many years,otherwise we would find it difficult tooperate. However, the situation is alwaysfluid because of new consumer devicesand services, technical developments andchanges in business processes. Newstandards and additions to existing onesfrequently need to be progressed, andthe pace of change is accelerating.

In the early days standards were setnationally. For the past twenty years orso, we have seen a steady movementtowards international and global bodies.But there is still no dominant body withresponsibility for all the standards that areused in broadcasting, and indeed we nowfind there are standards from thetelecommunications and IT industries thatare equally relevant to our business.Thismeans we often have to follow workthrough several standards bodies to

ensure we have the right tools for us atthe end.

Because of our pro-active approach tostandards, engineers from BBC R&I arefrequently invited to join standardisationgroups, recognising not just their technicalcontribution, but the fair and reasonablebalance we are able to broker whendiscussions become fractious andcommercial. In many groups we are alsoasked to take the chair.

There are a number of otherstandardisation bodies and organisations,for example MPEG, the TV-AnytimeEuropean Users Group, and sometechnical groups in the EBU and SMPTE,where we hold membership orotherwise follow the proceedings, tomake sure the BBC’s interests are notbeing compromised.

Where possible we would like to seestandards that are:

OpenThe standard is fully described indocuments that are obtainable byanyone, so that any competent engineercan read the specification and recreatethe technology.

and

FreeOr at least available on reasonable terms.Even a small charge for each instance of astandard being used can add up, giventhe volume of the BBC’s output and sizeof our audiences.Worryingly, we arestarting to see instances of royalties beingdemanded after a standard has beencommitted to.

International

The BBC has always worked in a so-called horizontal market where many players can share the same

technology for different business purposes.The only way we can make this work is by adopting

standards. Standards define almost every part of the broadcast chain, from before the capture of images

and sound, all the way through to the display and loudspeakers at home.We therefore make a serious

effort to help standards bodies to create proposals which will work well for public service broadcasters

and in a complementary way to the commercial goals of private industry.

WO

RK

ING

WIT

H U

S

Digital Video Broadcasting Project(DVB)The Digital Video Broadcasting Project (DVB)is a consortium of over 260 companiesand other organisations developing globalstandards for the delivery of digitaltelevision and data services.Their workhas now extended beyond broadcastingto the convergence of Internet and mobilesystems in the home. DVB specificationsare offered to ETSI, CENELEC, the EBU,or the ITU to create thestandards themselves.

BBC R&I has been instrumental ina number of DVB standards.

DVB MHP IPTV specificationThe DVB Multimedia Home Platform(MHP) is an advanced platform forinteractive TV.This year DVB hascompleted an IPTV profile for the MHP.

We contributed and edited theTV-Anytime parts of the specification,based on our earlier work in the myTVand Share-it collaborative projects.This allows the MHP to provide richprogramme guides based on informationdelivered via broadband IP.

DVB-CPT The first elements of the DVB CopyProtection & Copy Management (DVB-CPCM) were released in November 2005as the DVB ‘Blue Book’.This includes theDVB CPCM Reference Model and UsageState Information (USI) – an interoperabledescription format for the copy controlconditions associated withprotected content.

DVB-GBSThis group defines the carriage and sig-nalling of data and SI (including TV-Anytime)in the DVB transport stream.

DVB-SUBWe chair this group on subtitling, forwhich we recently rewrote the relevantstandard to enable subtitles atHD resolutions.

DVB-T2We lead the technical study group,developing an updated specification forDVB-T to improve transmission efficiencyand ruggedness.

ID3.orgID3.org is a group which has been definingmethods of tagging audio files withdescriptive information or metadata.We have developed an addendum to theID3 specification for the visually impaired,that allows an audio description of thetextual information to be hidden insidethe main audio file.

The Digital Media Project(www.dmpf.org)The Digital Media Project (DMP) is anon-profit Association. Its mission is topromote the successful development,deployment and use of digital media thatrespects the rights of creators and rightsholders to exploit their works, the wishof end users to fully enjoy the benefitsof digital media and the interests ofvalue-chain players to provide productsand services.

The DMP has wherever possible usedexisting technologies, for exampleMPEG-21, as the basis for a numberof primitive functions. It is finalising thethird release of its interoperable DRMplatform specification IDP-3.

SMPTE and the AAF AssociationThe Society of Motion Picture andTelevision Engineers is based in the USAand has produced many relevantstandards.We are active members of theSMPTE and the AAF Association, as theycontinue their work in refining standardsrelevant to television production. Ourinterest for this work lies mainly in the fileformats used for capture and storage ofcontent in tapeless production systems.

We are a participant in several SMPTEprojects: Reg-XML, which is developing anXML representation of MXF and AAF,SMPTE 410M, which is developing thebuilding blocks for incorporating otherdata, such as subtitles, in an MXF file andSMPTE 377M which is currently revisingthe core MXF standard.We are alsoworking on the SMPTE metadata elementsand groups registers which hold thedefinitions of the metadata carried in MXFand AAF.Additionally, we chair the SMPTEAAF Object Specification group, which istaking on the technical responsibility forthe AAF specification.

We are similarly closely involved with theAAF Association, having a member on theBoard of this trade body, and we lead itsengineering programme.The AAFAssociation has, to date, been the groupdeveloping the AAF standard but with thetechnical responsibility for this beingpassed to the SMPTE, its work is nowbecoming more user and workflowfocussed. Reflecting this change, the AAFAssociation has renamed itself as theAdvanced Media Workflow Association.

We have also been highly active in aSMPTE working group standardising theMedia Dispatch Protocol for coordinatingfile transfer of content betweenproduction centres over IP networks,including the Internet.The group has madegood progress, and this work will result inan important standard allowing the BBCto share content with its productionpartners in an open way.

Our video coding technique Dirac is alsobeing standardised within the SMPTE.

Audio Engineering Society (AES)Most of the digital interfaces used in theaudio industry are based on standardsmanaged by the Audio Engineering Society.

We have contributed to two standardsthat have recently been ratified. AES 51supports delivery of ATM cells overethernet. AES 52 puts unique identifiersinto AES 3 transport streams, so that

media managers can handle them muchas they do pre-recorded audio files,automatically tying them to metadata forexample.AES 52 also opens the possibilityof automatic routing.

AES 47, the audio-over-ATM standardextensively used in the BBC, has recentlybeen re-ratified following a routine five-year review.

WorldDMB Forum andDRM ConsortiumThere are many new platforms to deliverradio content that was once the preserveof analogue AM and FM technologies.The emerging technologies, such as DRMand DAB+, are seen as competitive bysome and as complementary by othersto our existing digital platforms. Ourinternational presence in the world canalso create conflicts between thedomestic agendas of different countries.By engaging with a number of internationalfora, such as the WorldDMB Forum(formally WorldDAB), the DRMConsortium and conferences acrossEurope,Asia and the Americas, we havebeen able to put forward the positive andpromote a reasoned co-operationbetween technologies that support ourcore values.This has ensured ourcontinued success in bringing clarity tothe messages delivered to our listenersfrom the companies that design and selldigital radios in all our markets.

Institute of Electrical andElectronics Engineers (IEEE)The IEEE has a wide range of standardsunder its remit.The main areas of interestto us are the wired and wirelessnetworking standards, where the IEEEhas become the body preferred to ETSI.

We are participating in the IEEE 802.11standardisation process, with a particularinterest in IEEE 802.11n, whose ratificationis now not expected until late next year.We hope to influence the work so thatnew standards will enhance the potentialof WiFi for video production.

http://www.dmpf.org/

W3CThe World Wide Web Consortium(W3C) develops interoperable tech-nologies to lead the web to its fullpotential.We are members of the‘Timed-Text’ working group.

International ElectrotechnicalCommitteeThe IEC prepares standards for electrical,electronic and related technologies. Insome respects it overlaps work in ETSI,IEEE and CENELEC, among others.Usefully, it seeks to lay down performancebounds, for example, the sensitivity onecan expect from certain types of radioreceiver, to ensure that systemswork effectively.

CENELECCENELEC, the European Committee forElectrotechnical Standardisation preparesvoluntary standards for electrical andelectronic goods and services.The mostimmediate relevance to our work istheir influence on the specificationsof receiver performance.


BBC Information & ArchivesStandards continued

WO

RK

ING

WIT

H U

S

A significant part of I&A’s project work isdone in collaboration with otherorganisations:

Spoken WordSpoken Word is a five-year DigitalLibraries Content project that startedin August 2003. It is jointly sponsoredby the UK Joint Information SystemsCommittee and the US National ScienceFoundation.The purpose is to allow BBCradio archive content (and other audio)to be accessible online to UK universitystudents. It is the only direct mechanismto get archive content into the highereducational sector, and has given usvaluable experience of ‘digital repositories’and associated access technology, all ofwhich is useful for BBC’s iPlayer and forOpen Archive initiatives. The otherpartners are the Glasgow CaledonianUniversity, Northwestern University,Chicago, and Michigan State University.

PrestoSpacePrestoSpace started in February 2004and has just passed its third review. It isa large EU Integrated Project with 38partners, running for four years. It is a‘public value’ project; the BBC and itstechnology partners know how topreserve audiovisual content efficiently,using a factory approach (for examplethe BBC Archive Radio Digitisationproject at Maida Vale). Our knowledgeand standing in audiovisual preservationmeans that the rest of Europe’saudiovisual collections look to us forguidance. PrestoSpace is the mechanismwhereby the knowledge and approach ofBBC and partners can be made available

to all audiovisual archives – to makepreservation work ‘better, fasterand cheaper’.

A number of other collaborations haverecently begun, part-funded by either theDTI or the European Commission.

AVATARAVATAR is part funded by the DTI.It is defining IT storage solutionsoptimised for audio visual contentand data from oil and gas exploration.It began in November 2006 and finishesin 2009.

ESTASTARECTASTAR is part funded by the DTI.It is devising architectures to accessexabytes of storage at terabit/s rates.It began in November 2006 and runsfor two years.

SEMEDIASEMEDIA is part funded by the EU.It will develop techniques, environmentsand tools for media labelling, searchingand retrieval from very large collectionsof heterogeneous data. It started inJanuary 2007 and finishes in 2009.

Video activeVideo Active is part funded by theEU. It is putting historical and heritagetelevision material onto a shared, publicwebsite, and providing tools to supportfinding material in different languages.It started in September 2006 and runsfor three years.

Other I&A projects are:

Computer AssistedIndexingComputer Assisted Indexing is a semiautomated indexing and taxonomymanagement tool for cataloguing andretrieving BBC archive material.

Off Air ComplianceRecordingOff Air Compliance Recording isthe replacement of the VHS-basedsystem for legal compliance recordingof BBC TV transmissions with anautomated system that will also captureinteractive elements (from a conceptdeveloped by BBC R&I).

Radio Digital ArchiveRadio Digital Archive is the capturing ofselected BBC radio output and relatedmetadata, with the ability to search andretrieve as digital files.

Online OrderingDevelopment of new systems for onlineordering of TV viewing material andmusic within the BBC, and for deliveringcommercial music direct to BBC users.

BBC Information and Archives (I&A) became part of FM&T in January 2007.

The project team working within I&A is responsible for several internal delivery projects, with a forecast

capital expenditure for 2007-8 of some £2.4 M.These projects are closely linked to the BBC’s Digital

Media Initiative, the creation of processes and technology that will allow the BBC to produce and deliver

content entirely digitally and without the use of tape.There can be no effective use of tapeless content

across the BBC without a digital archive at the centre.

Digital TV Group (DTG)The Digital TV Group is an industryassociation for digital television in theUK. This group defined the profile ofthe DVB standards which are now usedfor all Freeview services, and has alaboratory which is able to test complianceof consumer equipment with thosestandards. BBC R&I engineers haveworked closely with the Digital TVGroup to set the standards.

Now the group is concentrating on highdefinition television, radio spectrumissues, mobile TV, and domestic systems.

European TelecommunicationsStandards Institute (ETSI)ETSI is the dominant European bodywith responsibility for broadcasting andtelecommunication standards. Most ofthe broadcast standards we use arestandardised in ETSI, including DAB, theDVB suite of standards, and TV-Anytime.

European Broadcasting Union(EBU)The EBU is no longer the prime body forcreating broadcast standards as it was inthe days of MAC and NICAM. It has nowchanged its role to act as a consensus-building body, ensuring that public servicebroadcasters in Europe are able to developa common viewpoint. Much of the EBU’sformal documentation is published andthen referenced in other standards.

The EBU organises this work into a seriesof technical groups, and we contribute toall of these.

EBU Group B/MAEWe have recently taken part in a seriesof subjective quality tests organised by thisgroup, which evaluates multi-channel audiocoding technologies.

EBU Group N/ACIPSet up early in 2006, this group isdeveloping an interoperability standardfor equipment for audio contributionsover IP, in cooperation with manufacturers.It is building a test engine for verifyingcommercial equipment against the

standard, and is also compiling a set ofrecommendations for operational practice.

EBU Group N/CNCSThis group, which we chair, has beenrunning for a number of years and is nowcontributing to an IEC group developinga common control standard for networkconnected broadcast and media production.

EBU Group N/VCIPThis is a new group set up at the start of2007, with similar aims to N/ACIP but forvideo contributions over IP.

EBU Group P/CPThis is an interdisciplinary group, analysingcommon processes in TV production,to answer the increasing demand foreconomical systems integration. It ismostly concerned with file-based pro-duction and its related metadata issues.

EBU Group P/DisplayWe chair this group which is addressingthe supply of professional TV monitors,used by all broadcasters to ensureconsistent picture quality and colorimetry.CRT (cathode ray tube) TVs are nowalmost unobtainable, and the new flatpanel technologies work in very differentways, not only from CRTs but also fromone another.There are parameters tospecify which simply did not exist withCRTs, or which could be taken for grantedbecause the CRT TVs all worked in thesame way.This group liaises closely withthe SMPTE and the UK DTG.

Europe National

The day-long event saw many interesting and varied ideaspresented, including proposals to mount a high qualitycamera on athletes, use a Nintendo Wii controller to operateproduction systems, and put wind turbines on BBC premises.An audience of around eighty technologists attended and usedthe optical voting system developed by the Production Magicteam to give immediate feedback after each presentation.Delegates also completed feedback forms for a more con-sidered view on the proposals. Following each Forum a morestructured review process takes place, with volunteersdiscussing each idea in depth. Ideas that have been presentedto the New Ideas Symposia, the forerunner to the Forum,are also considered alongside those from the Forums.

April 2007 will see the start of work to build a RapidPrototyping and Research Centre (R-PARC) at KingswoodWarren.This will be a wholly configurable space with genericequipment plus ‘per project’ specialist technologies, as requiredby the ideas flowing through the Centre.The aim is to spendshort but intense periods of time researching and developingideas, perhaps creating prototypes or demonstrators orinvestigating proofs of concept.

44 APRIL 2006 – MARCH 2007 45

The Future Media Innovation team aimsto provide the BBC with product andservice ideas from internal and externalinnovation networks. It aims to developand support wide ranging and efficientinnovation networks, and to drive newproducts into the business. Our currentwork is based on developing tools forengaging with innovators outside theorganisation, loosely grouped in foursectors – corporate partners, academia,independent digital media productioncompanies, and ‘lead users’. Here aresome details about projects aimed atthese sectors.

Participatewww.participateonline.co.uk

Participate explores convergence inpervasive, online and broadcast mediato create new kinds of mass-participatoryevents in which a broad cross-sectionof the public contributes to, as well asaccesses, contextual content – on themove, in public places, at school andat home.

Participate is a three-year collaborativeproject supported by the DTI andEPSRC blending expertise in onlineservices, ubiquitous computing, broadcastmedia, sensors, pervasive gaming andevents, and education. Our partners areBT, Microsoft Research Cambridge, BlastTheory, ScienceScope, University ofNottingham and the University of Bath.

The consortium is working together todevelop scalable solutions for managedevents and campaigns that engage andmotivate participants over sustainedperiods of time.We are developing toolsfor the public to author, share and discusscontent using their own devices, and forprofessionals and experts to collate andedit contributions for publication overbroadcast and interactive channels.

The BBC is working with a range ofpartners to develop a series of trialsand events based on the theme of ‘theenvironment’ capturing and contributing

information about their local environ-ment to build a national picture acrossthe UK.

Arts & Humanities ResearchCouncil (AHRC) PartnershipThe aim of this partnership has been todiscover whether there was scope for along term partnership between our twocommunities, if there was a demand forenabling effective knowledge transfer anddelivering mutually beneficial research,and how that research could turn intojoint projects with a mutual strategic fit.The response after three initial summitswas overwhelmingly positive and showedthat a specialist arts and humanitiesacademic community which has a deepunderstanding of the drivers andinhibitors of change is an untappedresource for insight, expertise, and ideasfor BBC FM&T.

After the successful summits (documentedat http://tell.memore.info) a joint fundingcall for collaborative projects wasannounced in January 2007. Over 60applications were received and 26 madeit to a shortlist. Five of these have nowbeen chosen to go forward for jointfunding, giving clear indications that thereis a strategic fit for both communities.

Once the projects have begun we willbe monitoring progress, disseminatingoutcomes and measuring success andimpact on the business within bothcommunities. This type of partnershipis a first for the BBC and the AHRCand, should it prove successful, therewill be further funding calls. Meanwhile,we are developing a sophisticated onlinebrokerage tool which will track all theexisting relationships as well as helpingto develop new ones. We will be holdingmore summits in autumn 2007.

BBC BackstageBackstage.bbc.co.uk is the developer/designer network from the BBC. It is anopportunity for the BBC to engage withthe external developer community, offersome of the data and services the BBC

produces, and share them with thirdparty, non-commercial developers.

Over the last year we have gone outto developer communities all over thecountry in a series of meet-ups, eventsand university tours. We have establishedpartnerships with O’Reilly,Yahoo!,The Obvious and the Guardian.We’velaunched our new in-depth podcastsand started on a process to allow theBBC much more flexibility in regards todevelopment and prototyping into thenear future. Our next major project isthe Yahoo!/BBC Backstage ‘Hackday’ –a huge event involving over 400developers at Alexandra Palace on16 and 17 June, 2007 – more detailsat www.hackday.org

BBC Innovation Labshttp://open.bbc.co.uk/labs

BBC Innovation Labs is a process forworking with the independent mediaproduction companies on early-stageprototypes for new products andservices.The process has three stages:open days where BBC commissionersdescribe their strategic needs and thebriefs for the Labs; an open submissionprocess on the Labs site; and a five-dayresidential workshop where thecompanies work with BBC and externalmentors to develop their ideas.

After successful pilots in 2005/6, weworked with ten regional partners todeliver four Labs in 2006/7, coveringScotland, North England, South Englandand London. This year’s Labs werelaunched with 13 open days in citiesacross these regions, and we receivedover 500 proposals.These were filtereddown to 40 ideas that were developedover the five-day Labs, with 18 ideasbeing selected for further development.

Future Media InnovationThe Innovation Forum


WO

RK

ING

WIT

H U

S

In January 2007 we held our first FM&T Innovation Forum at the Science Museum’s Dana Centre.

The Forum will meet three times per year and is part of the innovation process that FM&T uses to

discover, develop and implement its most important innovative ideas. It acts as a focus to trigger and

socialise proposals as well as the start of the peer review process.

Surround video projectJigna ChandariaPaul DebenhamStephen JollyTim SargeantGraham ThomasBruce Weir.

Audio navigation visualisation projectAndrew Mason.

Augmented realityAdrian Woolard.

New Ideas Symposium teamMichael EvansMatt HammondStephen JollyAlia Sheikh.

Huw Williams hosted the event and also presented theinaugural Innovation Awards:

http://open.bbc.co.uk/labs/

http://backstage.bbc.co.uk/

http://www.participateonline.co.uk/

http://tell.memore.info

46 APRIL 2006 – MARCH 2007 47

IndexMore Information


MO

RE

IN

FO

RM

AT

ION

IND

EX

WWhhaatt ttoo ddoo iiff yyoouuwwaanntt ttoo kknnooww mmoorree..

This Annual Review has deliberatelybeen aimed at a broad readership.As a consequence, the articles areconsiderably shorter than they mightotherwise have been, and we haveavoided going into deep technical detail.

For those who wish to learn moreabout any of the projects, there areseveral places you can find more.We have moved away from paperdocumentation, and now rely on theInternet to deliver our output.Thismakes it much easier for everyoneto gain access to our work.You canfind details of our projectson our website.

WWeebbssiittee:: bbbbcc..ccoo..uukk//rrdd

Projects Many of the projects at BBC R&Ihave a special set of pages, either onour website, or hosted by others suchas our collaborative partners orSourceForge.The aim is to be openabout our work whenever possible.

White Papers We publish reports and tutorials aboutour work as a series of White Papers.

Our papers are published in placessuch as journals and conferences.We have a policy to retain copyrightwherever possible on these pub-lications.This allows us to offer theseworks as White Papers too, making allour output accessible in one place.

Lectures We receive many invitations to lecture.Our intention is to try to inform awide range of interested groups, fromthe high level professional conferencesand seminars, to small local specialistinterest groups.

Whenever the lecture is being givenin a public event, we try to publiciseit in advance through our website.

3G telephony 4 – 5, 11– 12, 23

AAAF (Advanced Authoring Format),

21–22, 41

AES (Audio Engineering Society), 36, 41

AHRC (Arts & Humanities Research

Council), 45

ARENA (EU Project), 27, 39

Arqiva, 32

Audio over IP, 31

AVATAR (DTI Project), 43

AXMEDIS (EU Project), 37, 39

BBAMZOOKi, 20

BARB (Broadcasters Audience Research

Board), 27

BBC Backstage, 14, 45

BBC Innovation Labs, 45

BBCi, 9, 26

CCabot Communications, 8

CALLAS (EU Project), 15, 39

DDAB, 4, 28, 32–33, 36, 42

DAB+, 28

Digital Switchover, (see Switchover)

Dirac, 34–35, 41

Dirac Pro, 34–35

DMP (Digital Media Project), 37, 41

DRM (Digital Radio Mondiale), 28, 41

DRM (Digital Rights Management), 37, 41

DTG, 7, 8–9, 13, 25, 42

DVB (Digital Video Broadcasting Project),

6, 9, 13, 26, 32– 33, 37, 40– 42

EEBU (European Broadcasting Union), 22,

31, 34, 36, 40, 42

ECTASTAR (DTI Project), 43

EICTA, 6, 25

ETSI, (European Telecommunications

Standards Institute) 9, 28, 40, 40– 42

FFloorman, 22

free-d, 16

Freesat, 5, 26

Freeview, 4–5, 6, 8–9, 13, 26, 42

GGames consoles, 4, 14, 29

HHD (High Definition TV), 4–5, 6–7, 9,

16–7, 23, 26, 30, 33, 34–35, 38, 4–42

IIBC 2006, 5, 8, 16, 22, 32, 35, 39

IEEE, (Institue of Electrical and Electronics

Engineers) 41–42

iMP, 11

iPlayer, 4–5, 10–11, 22, 34, 43

iview, (DTI Project), 17, 39

JJFMG Ltd, 23

KKamaelia, 29

MMATRIS, 16, 39

MHEG, 9

MHEG-5, 8

MIMO, 23, 32–33

MixTV, 16

MPEG Audio Layer II, 36

MXF (Material Exchange Format), 19,

20–22, 41

NNAB 2007, 5, 35

National Grid Wireless, 28, 32

OOFCOM, 19, 23, 32

PP2 memory cards, 21–22

Participate (DTI Project), 15, 22, 37, 39,

45

Piero, 5, 9, 16

PMSE (Programme Making and Special

Events), 23, 33

PrestoSpace (EU Project), 43

PRISM (DTI Project), 30, 39

RRadio Digital Archive, 43

Radio Player, 10

Red Bee Media, 6, 11, 16

SSAMBA, 20–21

Schrödinger, 34

SEMEDIA (EU Project), 43

Siemens, 6, 11, 13, 24, 26

SMPTE (Society of Motion Picture and

Television Engineers), 35, 40, 41–42

Spoken Word (DLC Project), 43

Subtitles, 6, 22, 27, 41

Surround video, 18, 44

Switchover (Digital Switchover) 4–5, 13,

15, 24–25, 38

TTV-Anytime, 11, 13, 14, 40, 42

UUser generated content, 15

VVideo Active (EU Project), 43

WW3C, 22, 42

Web 2.0, 5

WiFi, 31, 41

WiMax, 23, 31

WorldDMB, 28, 41

48 APRIL 2006 – MARCH 2007

BBC Research & Innovation

Kingswood Warren

TADWORTH

Surrey

KT20 6NP

Website: bbc.co.uk/rd

Tel: 01737 839500

Fax: 01737 839501

© BBC FM&T 2007

Edited by Richard Marsden and Graham Thomas

Designed by Kevin Claydon and Keren Greene

BBC R&D Annual Review...

Documents

Transcript of BBC R&D Annual Review...