Advice on spectrum usage, HDTV and MPEG-4

Transcription

1 Advice on spectrum usage, HDTV and MPEG-4 Part of the HDTV Public Value Test This document is prepared for BBC Trust. Jeremy Klein, Mike Reynolds and David Johnston 23 August 2007 Ref BB026

2 Executive Summary This report addresses a series of mainly technical questions raised by the BBC Trust as part of the Public Value Test (PVT) being applied to the BBC Executive s HDTV proposal. Our work started in early June 2007 and was originally substantially complete by early July However in mid-july we were asked to comment on a proposal by Ofcom on how HDTV could be provided for, and to update the report accordingly. While we have included a commentary in the report, HDTV is a live issue for the BBC, Ofcom, the Government and others, and is certain to change further during the period of the PVT. This report does not take account of developments beyond mid-july The BBC Executive s HDTV proposal is not to convert all its output to high definition, it is to introduce a single HD channel. On DTT this would be either 4 hours overnight, 9 hours during the day, or a combination of both. However, there is a current debate over whether Ofcom should make specific allowance for HD in DTT spectrum, particularly that spectrum released by switchover. After Ofcom initially indicated its intention not to make any specific allowance, the HD for All campaign was launched to lobby for the main channels to be made available in HD on Freeview. Occurring in this context, the BBC Executive s HD proposal is being seen by many as the first step in a bigger process to convert a significant proportion of its output to HD, and therefore a flagship decision. The technical questions go beyond the initial feasibility of the Executive s HDTV proposal to address longer term questions over HDTV. The questions raised by the Trust relate to two basic themes: (i) the capacity within currently allocated DTT spectrum to transmit HDTV, and (ii) the dynamics of the potential uptake of HDTV and the implications for broadcasters and consumers. (i) HDTV Capacity DTT is transmitted from six Multiplexes, of which two are owned by the BBC. Between them the six Multiplexes have a capacity of 120 Mbps. At present there is virtually no spare capacity within the 120 Mbps, and certainly not enough capacity to transmit a full HDTV service alongside all existing services. However, by the time switchover is complete, four of these Multiplexes will have been upgraded from 16QAM to 64QAM. This is known as mode change and will make a total of 24 Mbps extra capacity available. This then holds out the possibility of using some capacity for HDTV. The report looks at the feasibility of doing this and the number of HDTV channels that could, in principle, be included. The number of channels depends on the bit-rate required for an HDTV channel, but this is not a fixed quantity. It varies according to the quality level required, and also on the efficiency of compression. The chosen compression standard for HDTV in Europe is MPEG-4 but this is a relatively new standard. The performance of MPEG-2, the standard used for SDTV, has improved by a factor of four over its lifetime. Accordingly, MPEG-4 is also expected to improve from its 2007 level. The extent of improvement is a key uncertainty, however. The report concludes that: i

3 Currently, satellite HDTV broadcasts to Europe use between 10 and 18 Mbps per channel (picture, sound and subtitling etc), and this is approximately the capacity that would be required on DTT In 2012, an HDTV MPEG-4 service will require between 6 and 13 Mbps depending on both the quality level required and the extent of improvement in MPEG-4 In 2017, an HDTV MPEG-4 service will require between 3.5 and 9.25 Mbps. Assuming a reasonable degree of improvement in MPEG-4 we expect HDTV to require between 6.5 and 11.5 Mbps in 2012 and between 4 and 7.5 Mbps in Given the BBC s emphasis on quality we would expect BBC channels to be operating at the upper ends of these ranges. In the absence of other calls on the capacity of the two BBC Multiplexes, after switchover there will be a total of 9 Mbps available across the two. If this were available on a single multiplex, then there would not be enough capacity for a single HDTV channel at present. However, around 2010 there will be just enough capacity for one HDTV channel. However, there would not be enough capacity for the high quality picture that the BBC Executive aims to transmit. Despite this, the channel could be launched and its quality could be allowed to improve over time as MPEG-4 coding becomes more efficient. Beyond this one channel there are a number of ways in which more capacity could be made available, and therefore more HDTV channels transmitted, within the existing DTT allocation. These include: 1 Reducing the bit-rate of current SD services with associated picture quality degradation 2 An alternative, more radical, mode change to a lower FEC rate 3 Switching SD services to MPEG-4 4 Reorganising channels between the Multiplexes, for example by moving some non- PSB services to the commercial Multiplexes 5 Switching transmitters from DVB-T to DVB-T2 Phase 1 6 Switching the five transmission channels used by one PSB Multiplex to give five single frequency network (SFN) Multiplexes. In each case there are benefits, but also disadvantages. These include: losing regional services by making some channels entirely national, reducing the number of homes covered, re-planning and reorganising the Multiplexes, and necessitating consumers to replace their existing receivers and/or TV aerials. Shortly before the project was planned to be completed, Ofcom proposed a scenario which is a combination of 1, 4 and 5 above. This scenario involves reorganising the content of the existing six Multiplexes onto five Multiplexes post switchover, thereby releasing one Multiplex for HDTV. By changing the transmission mode of this one Multiplex to DVB-T2 Ofcom envisages that approximately 32 Mbps would then be available, initially allowing three HDTV channels and rising soon after to four. It is our judgement that this scenario judiciously avoids some of the potential disadvantages of the options outlined above. The basic principle of freeing up a single multiplex by redistributing the others is sound because there is sufficient capacity to do this. However Transferring some channels from the PSB multiplexes to the commercial ones would change which channels a small proportion of households have available to them. The commercial Multiplexes currently reach fewer households than the PSB Multiplexes owing to power levels and other technical factors. After switchover, the ii

4 commercial Multiplexes will continue to reach fewer households, with the additional reason that they will utilise comparatively fewer transmitters than the PSB multiplexes. (It is currently envisaged that post-switchover the PSB Multiplexes will reach 98% of households whereas the non-psb Multiplexes will reach 90% of households.) Thus, moving channels from a PSB multiplex to a non-psb Multiplex will mean that some households will loose access to channels that they previously had access to. However, if the transfer was done simultaneously with switchover in a region, the small proportion of households that would be unable to receive these channels after switchover would also not have been able to receive them before switchover. They would therefore not experience the loss of a channel as such. Admittedly they would not get to receive Freeview channels they are currently unable to receive but are planned to be able to receive after switchover. This is because, after switchover, increased power levels will ensure that the coverage of a non-psb Multiplex will be greater after switchover than the coverage of a PSB Multiplex before switchover. The loss of access to some Freeview channels could be set against the gain in access to HDTV channels. This disadvantage could be reduced by encouraging or compelling the operators of non-psb Multiplexes to build them out further. The broadcasters may already have plans for how they would use the extra capacity delivered by mode change The proposal includes a reduction in bit rate of the BBC s SD output which would reduce quality. Ofcom has been optimistic in the timescale for the introduction of DVB-T2, which according to the plan would be deployed in early DVB-T2 is expected to be a draft standard in A safer assumption would be that both transmission equipment and consumer reception equipment for DVB-T2 would be available in late By tying HDTV to DVB-T2 there is a risk that HDTV will be delayed. Ofcom has been optimistic in the bit rate improvement of DVB-T2. The objective behind the standardisation work is to achieve at least a 30% improvement with minimal reduction in coverage. Ofcom has made no reference to coverage reduction. Ofcom assumes that just three HDTV channels would be launched in the HDmultiplex in 2009, rising to four in The bit rates may not be consistent with the BBC Executive s aspirations for HDTV in terms of quality at least initially. That said, it might be argued that the quality will be at least sufficient to get HDTV launched. Introducing DVB-T2 could potentially create compatibility issues with receivers on rescans. If a receiver finds a DVB-T2 channel that it cannot properly decode, it should skip over it. This is the response of receivers to analogue channels at present. However, this response is untested with DVB-T2 and a few existing receivers may crash or hang if they receive a DVB-T2 Multiplex but are unable to decode it. The scale of this potential problem cannot be assessed for certain at this stage in the absence of DVB-T2 signals but it is not likely to be very significant. If necessary, some equipment may be upgradeable through over-air downloads. It would nonetheless be advisable to conduct receiver tests once DVB-T2 signals are iii

5 available. Tolerance to DVB-T2 should be added into the D-Book specification for receivers as soon as is practicable. In summary we conclude that, at a technical level, between one and four HDTV channels could realistically be provided within the existing DTT allocation through six Multiplexes. However, reorganising channels across the six Multiplexes would need coordination between the Multiplex operators, and there can be no guarantee that the necessary commercial agreements could be obtained. (ii) Uptake and implications of HDTV There is an evident momentum behind HDTV fuelled by large flat panel HD ready screens, the availability of HDTV on other platforms, most notably satellite, and the transition to HD in recorded media and games consoles. It is likely that consumers expectations will change over time to see HD as the norm. Unless DTT becomes HD, consumers are therefore likely to abandon DTT in favour of satellite or cable. The BBC has a special duty to achieve universality. Considering only fixed TV sets in the home, delivering HDTV via satellite could achieve a similar level of universality as DTT. If portable reception is required to be in HD, DTT is undoubtedly more universal. Opinions differ on the need for portable reception to be in HD. If portable sets were not to require HD, then one scenario would be for HDTV to be provided by satellite for fixed sets and SD to be provided by DTT for portable sets. This would be a rational and economically efficient partitioning of spectrum. Since a new receiver is required for HD anyway forcing this receiver to be converted to be a satellite receiver would not be detrimental to existing terrestrial users. If the UK were only to provide HD on satellite it would be at odds with the majority of countries such as the US, Japan, Spain, Australia, France and Italy that have all made commitments to providing HDTV on the terrestrial platform. The UK currently has a strong consumer electronics industry and the trade association Intellect has expressed fears that the UK could lose out commercially if it were not able to continue to develop and sell DTT equipment into its home market. Given that one of the rationales for digital switchover was to retain and enhance the UK s lead in TV technology, it would be ironic to erode that lead in a transition to HD. In the longer term, to make a substantial transition to HDTV will require more capacity to be squeezed from the allocated DTT spectrum. If all six multiplexes were converted to DVB-T2, the total capacity would be around 190 Mbps. If all output were converted to MPEG-4, there could be space for around 30 HDTV channels in In practice there are radio channels and interactive services, so not all 30 would ever be available. This will entail the conversion of all receivers to MPEG-4, even for SDTV; and also, eventually, a transition from DVB-T to DVB-T2. If such transitions are likely, then there need to be plans to stop older generation equipment being sold. Alongside, consumers also need to be educated that TV equipment will continue to evolve just as PCs do - and that products will generally need to be replaced at intervals of no more than ten years. A complicating factor is that in the case of TV there are two logically separate areas of progress compression standards (eg MPEG-4) and modulation standards (eg DVB-T2). Unfortunately the two iv

6 domains are not synchronised in their respective standardisation processes and MPEG-4 is several years ahead of DVB-T2. This could lead to some consumers changing from MPEG-2 to MPEG-4 only to find that three or four years later they need to change from DVB-T to DVB-T2. The Ofcom proposal has the benefit that it pulls these two transitions into synchrony. v

7 Contents 1 Introduction Technical briefing MPEG-2, MPEG-4 and the bit rate of HDTV Compression technology The evolution of compression technology The longevity of MPEG Alternatives to MPEG MPEG-4 versus MPEG-2 as the codec standard for DTT Compression in practice The issues and trade-offs in setting compression rates Strategies for setting bit rates Current compression rates advocated or used for HDTV transmission on DTT Current compression rates advocated or used for HDTV transmission on satellite and cable How multiplexes are split Forecasting bit rates for HDTV Capacity available on the DTT multiplexes The capacity after digital switchover for the provision of HDTV in the allocation of spectrum currently envisaged for PSBs Mode change Moving Five and S4C/Tele G Additional technology-related gains from switchover Summary of the position on the BBC multiplexes post switchover Other options to create extra capacity Conclusions The move towards HDTV Will television become high definition? How can HDTV be provided? DTT Satellite Cable IPTV and DSL Does DTT need to become high definition? The need for a national strategy Switching DTT to high definition the broadcasting perspective The scope for introducing one BBC HDTV channel in the BBC s currently envisaged allocation One-off events in HDTV Accommodating five PSB HDTV channels A longer term view Switching DTT to high definition the consumer equipment perspective The importance of converting consumer equipment to MPEG Encouraging the early uptake of MPEG-4 equipment Equipment-led scenario Broadcaster-led scenario Dealing with the legacy of MPEG-2 equipment Non-intervention Managed switchover The interaction of MPEG-4 and DVB-T Appendix A: Terms of reference for advice to BBC Trust on spectrum usage vi

8 1 Introduction This report addresses a series of mainly technical questions raised by the BBC Trust as part of the Public Value Test (PVT) being applied to the BBC s HDTV proposal. Our work started in early June 2007 and was originally substantially complete by early July However in mid-july we were asked to comment on a proposal by Ofcom on how HDTV could be provided for, and to update the report accordingly. While we have included a commentary in the report, HDTV is a live issue for the BBC, Ofcom, the Government and others, and is certain to change further during the period of the PVT. This report does not take account of developments beyond mid-july It is pertinent to recognise that the BBC Executive s HDTV proposal is not to convert all its output to high definition, it is merely to introduce a single HD channel. On DTT this would be either 4 hours overnight, 9 hours during the day, or a combination of both. However, there is a current debate on whether Ofcom should make specific allowance for HD in the spectrum released by switchover. After Ofcom initially indicated its intention not to make any specific allowance, the HD for All campaign was launched to lobby for the main channels to be made available in HD on Freeview. Occurring in this context, the BBC Executive s HD proposal is being seen by many as the first step in a bigger process to convert a significant proportion of its output to HD, and therefore a flagship decision. In our view, while such an outcome could follow, it is not the inevitable consequence of the BBC Executive s HDTV proposal. Equally, not launching a service at this stage would not preclude doing so later. These questions as they were put to us originally are included in Appendix A. They relate both to the BBC Executive s specific proposals, and also to the wider debate on the future of HD. There are two basic themes: (i) the capacity within allocated DTT spectrum to transmit HDTV, and (ii) the dynamics of the potential uptake of HDTV and the implications for broadcasters and consumers. The themes are introduced below. (i) HDTV Capacity Questions 1, 2, 3, 5, 1 st part of 6, 1 st part of 7, 8 and 9 all relate to the issue of how many HDTV channels, if any, could be transmitted in the spectrum currently allocated to the BBC and other PSB broadcasters for DTT. How many HDTV channels could be transmitted? is a simple enough question to ask but is surprisingly difficult to answer. The reason is that so many of the basic parameters are not yet fixed. It is rather like asking how many glasses of wine can be poured from one bottle if neither the size of a glass nor the size of a bottle is known for certain. Specifically: The amount of space (known as bit rate) taken up by an HDTV channel is not fixed. It depends on the quality of transmission being sought, the compression standard being used (MPEG-2 or MPEG-4), the maturity of the compression standard, and whether the compression is being done in real-time or off-line. These issues are explored in Chapter 3. The amount of space available on digital terrestrial depends on the way in which the channels are put together in multiplexes, and on the way the signal is transmitted. The current situation does not allow for widespread HDTV. There are, 1

9 however, ways of freeing-up spectrum which could be used. The likely availability of capacity and ways to increase it is explored in Chapter 4. (ii) Uptake and implications of HDTV Questions 4, 2 nd part of 6, 2 nd part of 7 and 10 all relate to the ways in which consumers will respond to the HDTV proposition and associated approaches to ensuring that people have the right equipment. The move toward HDTV is explored in Chapter 5, together with whether it needs to be broadcast on DTT. Assuming that it is, Chapter 6 looks at the transition from a broadcaster perspective and Chapter 7 looks at the consumer equipment angle. To help the understanding of this document, we have included a brief technical briefing in Chapter 2. 2

10 2 Technical briefing Multiplexes In digital television, each transmitter broadcasts one or more multiplex, or mux. The amount of spectrum allowed for each multiplex is fixed as an 8MHz block of UHF spectrum, the same as used for one analogue TV channel. Several digital TV channels can be combined into a multiplex using different forms of multiplexing. Statistical multiplexing allows channels to be most densely packed. Transmission mode The total amount of data (the bit rate) that can be carried in a multiplex is not fixed, however, and depends on the transmission mode. The DVB-T standard to which current digital broadcasters adhere defines a number of transmission modes. Each of these modes has a different trade-off of coverage, capacity and resilience to interference. At present some of the multiplexes use a16 QAM mode but these will change to a 64 QAM mode at switchover. The 64 QAM mode will give these multiplexes more capacity after switchover. There are several other possibilities to increase the bit rate on a multiplex including a new version of the transmission standard, namely an embryonic DVB-T2. This standard will allow new transmission modes, with even greater capacity, for example there may be a 128 QAM mode. Full DVB-T2 signals will not be able to be decoded by existing DVB-T receivers, inevitably requiring receiver equipment to be replaced. It looks unlikely that DVB-T and DVB-T2 will be able to be mixed on an individual multiplex, making it difficult to achieve a gradual transition 1. MPEG Compression The amount of data (bit rate) necessary to encode a single TV channel depends on the way in which a picture is compressed. The compression standards in common use are MPEG-2 and MPEG-4. MPEG-2 is the older compression standard and it is used in the UK for standard definition (SD) television. The newer, more efficient, compression standard is MPEG-4. MPEG-4 will be used in the UK for high definition (HD) television on DTT and satellite. In the UK and Europe, MPEG-2 is assumed to be synonymous with standard definition and MPEG-4 with high definition. Technically, however, MPEG-2 is able to encode high definition, and is indeed used for this purpose in the UK on cable and in both the USA and Japan on DTT, though it is not an efficient strategy by modern standards. 1 A form of backward compatibility is a stated objective in the standard. One way would be to effectively mix DVB-T2 and DVB-T in a single multiplex. Though not impossible in theory, the technical feasibility of mixing DVB-T and DVB-T2 remains untested in practice. We understand that the BBC Executive takes the view that the two modes cannot be mixed and we accept this as a justifiable assumption. 3

11 Conversely, MPEG-4 is able to encode standard definition pictures, and is more efficient than MPEG-2. If the UK were to move to high definition broadcasts on DTT, it would do so using MPEG-4. Having moved to MPEG-4, both HD and SD quality pictures can be transmitted so broadcasters will not be locked into high definition quality on all channels. MPEG-4 receivers will be able to connect to both SD and HD displays. The bit rate for MPEG compression depends on the picture quality being sought. Invariably, the video signal takes the majority of the bit rate available. Composition of a channel Other data that may need to be transmitted alongside the picture include: sound Mbps service information (SI) Mbps interactive Mbps per service subtitles/audio description 0.1 Mbps. 1080i versus 720p Yet another complexity is the contest between 1080i and 720p. These are alternative ways of sending pictures - but they relate to display resolution too. 1080i is interlaced so it is more complicated to achieve good MPEG-4 encoding. None of the questions raised by the Trust concerns this debate but there are implications for the bit-rate. 4

12 3 MPEG-2, MPEG-4 and the bit rate of HDTV 3.1 Compression technology The evolution of compression technology It is widely believed that the future evolution of MPEG-4 is likely to follow a similar path as the historic evolution of MPEG-2. MPEG-2 MPEG-2 was first published by ISO as in The bit rates associated with MPEG-2 over time are shown in Figure 1 which is derived from material supplied by Tandberg, a company which supplies MPEG encoders. It can be seen that the bit rate fell almost by a factor of four in the ten years since the first generation MPEG encoder but is now starting to plateau. Figure 1. MPEG-2 performance since its launch in 1994 Source: Ludovic Copéré (2005) Strategic Outlooks for European High Definition and IP TV ( based on material supplied by Tandberg) MPEG-4 By MPEG-4 in this document we denote MPEG-4 Part 10 (also known as MPEG high profile level 4) which was first published by ISO in

13 MPEG-4 Part 10 should not be confused with MPEG-4 Part 2 which is aimed at mobile and IPTV services and does not support HDTV. MPEG-4 Part 2 was first published in This earlier standard is not relevant to the discussion of HDTV. MPEG-4 is in its early days and is limited to a large extent by the processing power and cost of encoders. The EBU suggests that using current technology 18 Mbps is required for imperceptible degradation of HDTV using MPEG-4 Part10 coding. It also notes that broadcasts generally operate at 60-80% of that (ie 10 to 15 Mbps for real-time encoded video). That is, broadcasters accept some level of quality loss and trade this off with other factors such as the number of channels. Note that audio and SI need to be added to this to obtain a full service. There is a lot of R&D being undertaken into MPEG-4 encoding and it is likely that this research will find its way into practice during the next few years. We believe that a prudent forecast is that MPEG-4 will achieve further substantial improvements, but not quite as dramatic as MPEG-2. There are two reasons for this: The fastest improvements in MPEG-2 were in its first few years, prior to publication by ISO. Because MPEG-4 has already been published by ISO, MPEG-4 is already sufficiently advanced that some early stage improvement has already occurred While the bit rates required in MPEG-2 improved by a factor of four in ten years, it can be argued that MPEG-4 already benefits from these improvements and there is less opportunity for MPEG-4 to achieve the same level of improvement in the future. This can be countered by observing that the improvements in MPEG-2 were not predictable at its adoption in Accordingly we estimate that MPEG-4 should see improvements by a factor of between two and three over the next 10 years. Post MPEG-4 The EBU has noted in a recent technical report that the lifetime of a video codec is about 10 years. By this they mean that codec technology advances enough in 10 years that broadcasters will accept the consequences of making a change. This timescale is about the time that it takes to replace a codec in the vast majority of receivers in the market. The upgrading of codecs is driven by commercial pressures to pack as much content as possible into the bandwidth available. There is activity in the research community on more advanced codecs, in particular a Scalable Video coding technique seen as a successor to MPEG-4 part 2. There is also research in distributed video coding. Much of the research is still based on more efficient encoding for MPEG-4-part 10. We are not aware of any other video codec standards in development that would replace MPEG-4 Part 10 although research continues, particularly in wavelet transforms a new technology likely to replace the cosine transform that underpins the current MPEG 2 and 4 codecs. Thus, even though there is no successor at the moment for MPEG-4, we can be reasonably confident that there will be a new standard in ten years time. This implies that we will be in a process of change for some time to come. By the time the UK can turn off MPEG-2, MPEG-4 will already be being superseded. In this process of successive generations of technologies superseding each other, the impetus for the introduction of new generations is largely commercial and the timing is 6

14 commercially determined. The timing of the phasing out of old generations of technology may in addition be influenced by government and regulators who recognise a duty to balance the commercial drive for improved performance with the need to offer a reasonable level of protection to those consumers who are still dependent on old technologies. Audio SD transmissions normally use MPEG1 layer 2 or 3 coded stereo audio. HD transmissions are generally expected use a surround sound audio, either Dolby AC3 or MEPG4-AAC. The latter is only occasionally used because it has little advantage over Dolby AC3. Currently a few European broadcasts use MPEG stereo audio with the remainder using AC3 surround sound. Whilst audio compression is also advancing, the move to surround sound is actually increasing the bit rate used for audio. DVB-T audio rates range from 128 to 256kbps. AC3 rates range from kbps. (The BBC trial is at the low end of that.) EBU TECH 3312 suggests 384kbps surround sound is appropriate for broadcast with HDTV The longevity of MPEG-4 MPEG-4 Part 10 is the variant of MPEG-4 used for coding broadcast TV pictures and in particular for HDTV. MPEG-4 is an extensible standard and so we would expect it to have a long life. The extensibility comes from it being a container for other standards. MPEG-4 Part 2 is older and is used for TV transmission to mobile handsets. We would expect a new more advanced codec to eventually arrive. This may be a new part to MPEG-4 or a completely new family. Note that the MPEG standards MPEG7 and MPEG-21are Multimedia standards and not video coding standards Alternatives to MPEG-4 Microsoft is promoting its VC-1 codec as an alternative and more advanced codec than MPEG-4 part 10. However, European broadcasters and equipment companies have already adopted MPEG-4 and we do not foresee a standards battle on the horizon as occurred between Betamax and VHS MPEG-4 versus MPEG-2 as the codec standard for DTT Given that MPEG-4 is the natural successor to MPEG-2, it might be assumed that wherever MPEG-2 is in use, it will be replaced by MPEG-4. This simple view of succession is complicated by the deployment of MPEG-2 for HDTV in both the USA and Japan. We therefore look at the territories separately. In the USA digital terrestrial television is well established with nearly 30 million HD MPEG-2 receivers in the market. Television has evolved differently, however, and a there is nothing like the spectrum scarcity found in the UK and most of Europe. There is far less pressure to constrain the bandwidth of individual channels, and as a result there is no pressure to change from MPEG-2 to MPEG-4. The reason for the different evolution is that universal coverage has never been a priority. Terrestrial TV has become centred on major conurbations with large gaps in coverage between conurbations. Each TV station owns and operates it own transmitter. Because analogue TV broadcast cannot occur on the adjacent channel because of receiver performance limitations, each transmitter has an adjacent channel unused. 7

15 The FCC has mandated switchover and awarded each TV station an additional channel to broadcast on in digital only. Post-switchover the spectrum will be reorganised and surplus spectrum will be returned to the FCC for auction. This has led to every TV station having 20Mbps of digital capacity. In most cases this capacity is being used to broadcast a single HDTV channel although some stations broadcast in SD only and a few have only a single SD channel in this capacity. The fact that there is still significant unused capacity in the US suggests there are no drivers to migrate MPEG-2 HDTV to MPEG-4 HDTV. In Japan, DTTV is less established with fewer DTT receivers in the market than the UK. Japan however has a history of fast uptake of technology so whilst technically it faces the same issues as the US we believe the cultural aspects make MPEG-4 adoption more likely. Most countries of Europe are adopting MPEG-4 for HDTV. Chipsets are starting to become available and we have had indications from chipset manufacturers that the latest generation of chipsets supports MPEG-4 in addition to MPEG-2. New receiver designs should therefore be capable of MPEG-4 decoding as long as they have suitable software. Since many receivers have upgradeable software they may well be field-upgradeable to support MPEG-4 even if they do not support it at the time of manufacture. The UK, being early into DTT, adopted MPEG-2 and has no current plans to change to MPEG-4. However the use of MPEG-4 is planned for HDTV. A key question is when and how MPEG-4 will replace MPEG-2. We believe the most likely scenario is that receivers will start to support both MPEG 2 and MPEG-4 decoding by default. This might happen within the next couple of years. The availability of MPEG-4 decoders will be driven by the economics of chip production, which in turn will be driven by standards used by broadcasters across Europe (including the UK). 3.2 Compression in practice The issues and trade-offs in setting compression rates MPEG compression involves an encoder at the transmitter end and a decoder at the receiver end. In simple terms, MPEG encoders seek to analyse what is happening to a picture in terms of the movement of objects and transmit this information in an efficient way. A more sophisticated encoder will achieve a greater compression (lower bit rate). The decoder then reconstructs a moving picture from the information about objects and their movement. The decoder is indifferent to the way in which the encoding has been done. Improvements in the performance of the decoder do not therefore have to match improvements in the performance of the encoder. The issues and trade-offs vis-à-vis compression are therefore entirely at the broadcaster s end. The tradeoffs are illustrated in the influence diagram of Figure 2. 8

16 Viewer satisfaction Cost of encoder Quality vs cost and capacity trade-off Quality Bit rate Mbps Encoding vs transmission trade-off Total cost Cost of broadcasting Figure 2. Influence diagram of factors related to compression rates The diagram represents a snapshot of the issues faced by a broadcaster. There are two basic trade-offs. Quality vs cost and capacity trade-off The quality of service drives both viewer satisfaction and bit rate. In the case of MPEG-type encoding, quality is the level of undesirable artefacts (unwanted patterning on screen). With current MPEG-4 encoding technology compression artefacts become noticeable on some content types at rates below 18 Mbps. Such artefacts are more visible on large screens than small ones. The trend towards larger screens for the main TV means that there will be continued consumer pressure on quality of service. However, high quality encoding entails higher bit rates. High bit rates then drive up the cost of broadcasting which is related to the amount of transmission capacity used. High bit rates also reduce the total number of channels that can be squeezed into a given block of spectrum. Broadcasters have to make this trade-off already in respect of standard definition broadcasts, and will have to do likewise with high definition. Different broadcasters make different judgements as to what levels of undesirable artefacts are acceptable to their customers, and also distinguish between the bit rates of different channels. For example, BBC1 currently uses 4.5 Mbps but the lowest rate BBC service we can identify is BBC parliament, which has not very demanding content, and operates in about 2.5 Mbps. The BBC has historically (for standard definition broadcasts) operated its services at relatively high bit rates. Commercial operators generally run at lower rates and some current commercial services operate in 2 Mbps, Encoding vs transmission trade-off The second major trade-off is between the cost of encoding versus the cost of transmission capacity. The job of the encoder is challenging in that it has to process vast amounts of information, normally in real time. The best performance encoders, which achieve greater compression for a given quality level, are more expensive. 9

17 Broadcasters therefore take account of the relative cost of transmission capacity when choosing the level of compression to employ. As discussed in section 3.2.4, the tradeoffs differ by platform. A further issue is whether the encoding is done in real-time or offline. Nikkei Electronics Asia -- July 2005, reported that the same picture quality as 5.5 Mbps realtime encoding could be achieved with 1.8 Mbps offline encoding, both for MPEG-4 Part 10. Broadcasters transmit enough of their material live that real-time encoding is the norm. On the other hand, video on demand services have no live output, so would typically be encoded offline in order to achieve better compression rates. The proposed BBC overnight HD service could potentially use this approach, though being overnight there may be little benefit in using it Strategies for setting bit rates There are two possible strategies to setting bit rates: Setting a bit rate and keeping it constant over time, thereby allowing the picture quality to improve as encoding technology improves. Setting a level of quality and then allowing the bit rate to reduce with improved encoding technology Each of these approaches has its risks. Low picture quality at start of an HD service could damage the reputation of HD in the critical launch period, but at least satisfies viewers appetite for variety of content. High quality at the start of an HD service obviously creates a wow factor, but implies few channels. The BBC s traditional emphasis on quality leads it naturally to the option of high quality and declining bit rate Current compression rates advocated or used for HDTV transmission on DTT HDTV MPEG-4 DTT data rates for the UK are shown in the table below. Note that all the figures in this section refer to the service bit rate, including video, audio and SI (service information) required for the service. 10

18 Table 1: HDTV DTT data rates for the UK Service/organisation Bit rate/ Mbps Status Notes BBC HDTV Trial 19.5/14.3 Trial service, active MPEG-4 BBC HDTV Service 15, reducing to 12 2 Advocated MPEG-4 Ofcom 8 Advocated MPEG-4 It can be seen that the BBC Executive is planning to drop its data rates over time as compression improves. Ofcom s estimate similarly takes account of improvement in compression. It will undoubtedly become possible for high quality compression to be achieved at 8 Mbps. The question in relation to Ofcom s estimate is the timescale by which this would occur. The equivalent data rates across Europe and the RoW are shown in the table below. HDTV services in the USA and Japan are transmitted in MPEG-2. Table 2: HDTV DTT data rates for Europe and RoW Country Bit rate/ Mbps Status Notes USA 19 Active MPEG-2 France 12, reducing to 8 Advocated MPEG-4 Germany 6-10 Advocated MPEG-4 Italy NA Trial MPEG-4 Norway Advocated MPEG-4 Spain NA Trial MPEG-4 Australia 13 Part time introductory service MPEG-2 Japan 14 Active MPEG-2 The BBC s HDTV trial is seen as the leading terrestrial broadcasting trial in Europe, with the majority of comment referring to the published results. This may be because other trials have not published results. 2 According to the Joint BBC Trust/Ofcom description of service, the BBC Executive expects that it might use spectrum capacity of up to 15 Mbps to provide the HD channel on DTT in the first instance but as compression technology improves, the BBC Executive says that it should be possible to carry the channel on DTT within 12 Mbps. 11

19 3.2.4 Current compression rates advocated or used for HDTV transmission on satellite and cable Technically, the transmission platform makes no difference to the level of compression required. However, as explained above, there is a commercial cost trade-off between transmission capacity and encoding cost. Encoders that achieve good quality at low bit rates cost more than ones that need a higher bit rate for the same picture quality. Satellite transmission capacity costs less than terrestrial transmission capacity. This results in the bit rates used on satellite being slightly higher on average than those on terrestrial. Virtually all European satellite services use rates of between 10 and 18 Mbps per service. It is noticeable that services targeted to the UK and Germany use higher bit rates than those targeted at France and Italy: see Table 3. This is unlikely to be an arbitrary difference. Different countries have, historically, aimed for different quality standards, and viewers expectations therefore differ. France also does not yet transmit in widescreen. Table 3: Current measured bit rates of European HDTV MPEG-4 satellite services Service Gross service bit rate (Mbps) Audio Bit rate (kbps) Satellite BBC HDTV 19.2/16 * 256(stereo)+384(AC3) Astra 28.2 (UK) Sky Sport HD (Stereo)+AC3 Sky Box Office HDTV AC3 Sky One HDTV 18.4/13 * 192(Stereo)+AC3 Discovery HDTV 16.4 AC3 Discovery HDTV 10 AC3 Hotbird 13 (France, Italy) Sky Cinema 10 AC3 National Geographic HDTV 10 AC3 Premiere HDTV 12.2 AC3 Astra 19.2 (Germany) Discovery HDTV 18 AC3 Sat1 HDTV (AC3) Source: * These services appear to have changed bit rate over the monitoring period Because cable systems are closed, there is less public information on their technologies. In the UK HDTV over cable is generally 18 Mbps MPEG-2 and we understand that it is similar for other countries. 12

20 3.2.5 How multiplexes are split When planning services, the mutiplexes tend to be split in a simple fashion to determine bit rates. For HDTV Satellite services this produces the current 18 Mbps rate. (2 services per multiplex) with a planned migration to 13 Mbps (3 services per multiplex), and potentially 9 Mbps (4 services per multiplex, although we have not yet heard of this rate actually proposed for a satellite service.) For DTT the capacity of a multiplex is lower, leading to proposals based around 12 Mbps per service. (2 per multiplex) with a migration to 8 Mbps (3 per multiplex). In reality though, a multiplex carries a broad range of different services to make best use of its capacity. This is particularly true of DTT where capacity is limited. It results in the capacity allocated to a particular service being that set by the broadcaster to meet their quality criteria. This is evident in DTT where current video rates vary between about 2.0 and 4.5 Mbps. 3.3 Forecasting bit rates for HDTV To work out the number of HDTV channels possible it is necessary to forecast bit rates for HDTV. We start with the situation in mid-2007 and then predict forward to 2012 and then 2017, accepting that predicting bit rates is not an exact science situation As discussed in section satellite HDTV broadcasts to Europe use between 10 and 18 Mbps per channel (picture, sound and subtitling etc). We have cross-checked this against a calculation based on current SDTV bit rates. In this approach we start with the bit rate for an SDTV MPEG-2 picture in We then multiply by a factor to take account of the relative improvement of MPEG-4 over MPEG-2 in This is assumed to be 1. We then multiply by a factor of 4 because an HDTV picture contains about four times the information of an SDTV picture. To the result we add-in the audio and other overheads which we estimate at between 250 kbps and 500 kbps 3. Current SD pictures range from 2 to 4 Mbps. Using the above method an HD MPEG-4 channel should range from 8.25 Mbps to 16.5 Mbps. This calculation validates the observed bit rates for satellite. Forecast to 2012 and 2017 MPEG-4 encoding will improve over time so as to reduce these bit rates. Using MPEG-2 as a base as discussed in section we predict that MPEG-4 will improve by a factor of between two (worst case) and three times (best case) over ten years. For the purposes of this report we have assumed a base case improvement factor of 2.5. The improvements to MPEG-4 apply only to the picture aspects of the channel; the audio and subtitling and audio description is not expected to change very much. To estimate the range we have assumed 250 kbps for this part of the signal in the low quality case and 500 kbps in the high quality case. 3 Extra information includes subtitling, audio description, interactive content, EPG and service information 13

21 Assuming constant ratio year-on-year improvement as with MPEG-2, Table 4 presents our predictions. Table 4: Predictions of bit rates for HDTV channels High quality, worst case improvement High quality, base case improvement High quality, best case improvement Low quality, worst case improvement Low quality, base case improvement Low quality, best case improvement To summarise: In 2012 an HDTV MPEG-4 service will require between 6 and 13 Mbps depending on both the quality level required and the extent of improvement in MPEG-4 In 2017 an HDTV MPEG-4 service will require between 3.5 and 9.25 Mbps Assuming the base case improvement in MPEG-4 we expect HDTV to require between 6.5 and 11.5 Mbps in 2012 and between 4 and 7.5 Mbps in Given the BBC s emphasis on quality we would expect BBC channels to be operating at the upper end of these ranges. 14

22 4 Capacity available on the DTT multiplexes 4.1 The capacity after digital switchover for the provision of HDTV in the allocation of spectrum currently envisaged for PSBs At present there is no spare capacity. However, various changes being made as part of the switchover process will affect capacity. These are explained below Mode change Switchover will enable transmit power levels to be increased and this, in turn, will allow the transmission mode to change from 16 QAM to 64 QAM for the BBC and NGW multiplexes. This will provide an extra 6 Mbps per affected multiplex as described in Table 5. Table 5: Multiplex capacity pre- and post-switchover PSB Multiplexes Commercial Multiplexes BBC Muxes D3/4 Mux SDN NGW Multiplex 1 Multiplex B Multiplex 2 Multiplex A Multiplex C Multiplex D Pre-switchover (Current) 2x18 Mbps 24 Mbps 24 Mbps 2x18 Mbps Proposed post-switchover 2x24 Mbps 24 Mbps 24 Mbps 2x24 Mbps The gross effect of mode change will be an increase of 12 Mbps across the two BBC multiplexes and a similar increase on the commercial multiplexes Moving Five and S4C/Tele G There are a number of PSB services that are moving between multiplexes at switchover. The significant ones of these are Five and S4C/TeleG (S4C in Wales only TeleG in Scotland only) which will move from a commercial multiplex to a BBC multiplex. There are other proposed changes but these affect only small amounts of capacity and so are ignored here Additional technology-related gains from switchover There are some additional technology-related gains from switchover. There is an interaction between the statistical multiplexing process and the bit rate which has the effect of amplifying the gains made by mode change. In effect, the higher bit rate gives more statistical opportunity for statistical multiplexing to be efficient. To estimate the magnitude of this effect we note that the average bitrate per service on the ITV/C4 multiplex is about 2.9 Mbps compared with the 3.0 Mbps on the NGW multiplexes. We deduce that this gain comes predominantly from the greater proportional gain from statistical multiplexing on the 64 QAM ITV/C4 multiplex. Scaled up, this implies that mode change from 16 QAM to 64 QAM will give an additional increase in capacity of a little under 1 Mbps on each multiplex that undergoes the mode change. 15

23 In addition the BBC could obtain a further 1 Mbps by including BBC1 in the stat multiplex pool, as is already done on Multiplex 2 by ITV for ITV1. BBC1 is not currently included Summary of the position on the BBC multiplexes post switchover Table 6 shows the cumulative effect of changes to BBC Multiplex capacity. Table 6: Summary of changes to BBC Multiplex capacity post switchover Mode change 16 QAM to 64 QAM Move channels from commercial Multiplexes to BBC Multiplexes Statistical multiplexing 64 QAM Statistical multiplexing BBC1 Net increase +12 Mbps -6 Mbps +2 Mbps +1 Mbps 9 Mbps The Ofcom DDR suggests rather more capacity would be gained, however our analysis suggests that Ofcom is assuming the capacity will be filled with low bit rate services, the bit rate being about that currently used by the non-psb services on Multiplex 2 (2.5 Mbps). 4.2 Other options to create extra capacity We have considered a number of options which would create extra capacity so as to allow for the introduction of HD services. These options are summarised in Table 7. While all the options would create extra capacity they would also involve consequences such as: Losing regional services by making some channels entirely national Reducing the number of homes covered Requiring the multiplexes to be re-planned including requiring non-psb services on the BBC Multiplexes to move into commercial Multiplexes Necessitating consumers to replace all their existing receivers Necessitating consumers to replace their TV aerials. 16

24 Table 7: Further options for increasing capacity, together with their consequences Change PSB capacity created & No of HDTV Channels available Earliest likely introduction Partial loss of regional opt-out capability Reduction in number of homes covered without network re-plan Consequences Requirement for non PSB services to move into commercial Muxes Replacement of all existing receivers Replacement of existing TV aerials 1. Reduction in bit rate of current SD services with associated picture quality degradation 2. Alternative mode change to 64 QAM and lower FEC rate 12 Mbps 1 channel 18 Mbps 1-2 channels With DSO No No No No No With DSO No Yes No No No 3. Wholesale switch to MPEG-4 22 Mbps 2-3 channels With DSO No No No Yes No 4. Move non-psb services to Commercial multiplexes 18 Mbps 1-2 channels With DSO No No Yes No No 5. Switch to DVB-T2 Phase 1 28 Mbps 2-3 channels 4 years from now No Yes No Yes No 6. Switch spectrum used by one PSB Multiplex to Five SFN multiplexes 102 Mbps 9-12 channels 6 years from now Yes Yes No No Yes The six options are described in more detail below. 1 Reduction in bit rate of current BBC services to release capacity for HD. This would degrade the picture quality of the current MPEG-2 SD services although some of that degradation may be made up for by advances in MPEG-2 SD encoding technology. We have assumed a 1 Mbps reduction in each BBC service. 2 An alternative mode change to the one planned for digital switchover exists. This would be to reduce the guard interval and FEC rate 4 and so increase the capacity of each multiplex to nearly 30 Mbps. Guard interval and FEC rate are separate technical options within the DVB-T specification: Reducing the guard interval reduces the tolerance of the signal to multipath propagation and also increases the amount of data that can be transmitted. 4 Also known as coding rate 17

25 Whilst this cannot be compensated for by an increase in transmitter power it can be compensated for using additional relay stations where necessary. Reducing the FEC rate reduces the tolerance of the signal to noise and interference but increases the amount of data that can be transmitted. This can be compensated for by an increase in transmitter power. A consequence of this change would be a reduction in the percentage of households that could receive the signal. This in turn could be mitigated by increase in transmitter powers, but higher power means more signal interference. The network could be redesigned to avoid the interference, but this would be a major exercise and include negotiating with neighbouring countries. Making these changes on just the BBC-operated multiplexes could increase the gross capacity to 84 Mbps, leaving 18 Mbps available for additional HDTV channels. 3 All the existing SD channels on the PSB multiplexes could be switched to MPEG 4. Given current compression technology we estimate this would reduce the capacity used for TV services by 30%. With a gross PSB capacity post DSO of 72 Mbps, and about 75% of this used for TV services (rather than audio, radio or interactive services) this would release 16.2 Mbps of PSB capacity. Added to the 6 Mbps released by the DSO mode change gives approximately 22 Mbps available for HDTV services. 4 Another way of gaining capacity would be to reorganise the multiplexes so as to free-up space on the PSB multiplexes. At DSO there will be an additional 12 Mbps of capacity available to the commercial multiplexes from the mode change. This might be used to carry services currently on the PSB multiplexes (Multiplex 1, 2, B) but not classified as PSB services. As of July 2007 these services include ITV2, 3 & 4, E4, More 4 and Film 4+1. If such services were transferred to the commercial multiplexes then 6 channels of 3 Mbps each or 18 Mbps plus the 6 Mbps from DSO mode change would be available for HDTV. Note that transferring these services to the commercial multiplexes would have to displace some of the current commercial services. 5 New broadcast standards are always in development and DVB-T2 is the embryonic next generation of the DVB-T modulation standards currently used for DTT broadcasts. Our current understanding is that the first stage of DVB-T2 defines a mode of operation that will increase capacity by 30% with minimal reduction in coverage. Each of the existing BBC multiplexes could thereby increase by 7 to 8 Mbps. Assuming that adopting DVB-T2 across all the PSB multiplexes would create an additional 30% capacity, and adding the 6 Mbps already available from DSO Mode change, would make 28 Mbps available for HDTV broadcast. This approach would require all existing STBs and IDTVs to be replaced. If all six multiplexes were eventually converted to DVB-T2, the total capacity would be around 190 Mbps. Timing is an issue. DVB-T2 has three parts but only the first one is detailed enough to consider at present. This part commenced standardisation work just a month ago in June First trial technical standards are due to be available in early Once standards are agreed, the D-Book will need to be modified and DTG testing, compliance and conformance regimes will need to be devised and agreed. Then the whole manufacturing chain from the chipset suppliers to the 18

26 equipment providers will need to adjust. This will take a minimum of 18 months. Even taking an optimistic view we believe commercial service could not start before the end of 2010 because of the lead time required for transmission and reception equipment. (UK launched DTTV in 1998, to a final specification from 1995). 6 A further enhancement that would be very disruptive but might be considered is to replan the PSB capacity. Post switchover there will be 32 TV frequencies carrying 6 National multiplexes, approximately 5 frequencies per multiplex. These multiple frequencies are required to allow for regional services and to reduce the number of transmitter sites needed for universal coverage. Reallocating one block of five frequencies to a set of SFNs (single frequency networks) would allow the spectrum currently used for 1 multiplex to support five multiplexes giving up to 96 Mbps of additional capacity. Enough for up to 12 HDTV channels. There are however a number of issues with this approach. New receiving aerials would be required for the majority of homes. The new capacity as well as one of the existing multiplexes would have to be identical throughout the UK (but Multiplex B is already national) There would be co-ordination difficulties with neighbouring countries More transmitter sites would probably be needed. There would be significant cost to provide the upgraded network An approach similar to this is being used in France to provide additional capacity for HDTV. The Ofcom scenario An alternative scenario combining elements of options 1, 4 and 5 has been proposed by Ofcom. This has a number of components: 1 Moving services. The gross capacity of all six multiplexes pre-switchover is 120 Mbps. This content can be redistributed into just five multiplexes post-switchover because the capacity would be the same (5 * 24 Mbps) 5. The effect of the redistribution is to free-up one complete multiplex. 2 Reducing SD bit rate. Ofcom also proposes to reduce the capacity allocated to the BBC s existing services from 36 Mbps down to 29 Mbps. 3 DVB-T2. The freed-up multiplex is then switched to the DVB-T2 standard. This increases the capacity of the multiplex to approximately 32Mbps, allowing the transmission of three to four 4 HDTV services. We have assessed this scenario from a technical perspective and believe it is broadly feasible. However we have some detailed comments: The basic principle of freeing up a single multiplex by redistributing the others is sound because there is sufficient capacity. Transferring some channels from the PSB multiplexes to the commercial ones would change which channels a small proportion of households have available to them. The commercial Multiplexes currently reach fewer households than the PSB Multiplexes owing to power levels and other technical factors. After switchover, the 5 The capacity on the five multiplexes will be slightly higher because the stat mux pool will have increased on each multiplex 19

27 commercial Multiplexes will continue to reach fewer households, with the additional reason that they will utilise comparatively fewer transmitters than the PSB multiplexes. (It is currently envisaged that post-switchover the PSB Multiplexes will reach 98% of households whereas the non-psb Multiplexes will reach 90% of households.) Thus, moving channels from a PSB multiplex to a non-psb Multiplex will mean that some households will loose access to channels that they previously had access to. However, if the transfer was done simultaneously with switchover in a region, the small proportion of households that would be unable to receive these channels after switchover would also not have been able to receive them before switchover. They would therefore not experience the loss of a channel as such. Admittedly they would not get to receive Freeview channels they are currently unable to receive but are planned to be able to receive after switchover. This is because, after switchover, increased power levels will ensure that the coverage of a non-psb Multiplex will be greater after switchover than the coverage of a PSB Multiplex before switchover. The loss of access to some Freeview channels could be set against the gain in access to HDTV channels. This disadvantage could be reduced by encouraging or compelling the operators of non-psb multiplexes to build them out further. The broadcasters may already have plans for how they would use the extra capacity delivered by mode change. For the BBC this includes more interactive content and extending the time over which some channels are broadcast. These plans may not have been taken into account in Ofcom s proposals. The reduction in bit rate of the BBC s output would reduce quality. The reduction of bit rate is in not strictly necessary to perform the redistribution of services, though it would have the effect of creating further capacity. Ofcom has been optimistic in the timescale for the introduction of DVB-T2, which according to the plan would be deployed in early DVB-T2 is expected to be a draft standard in A safer assumption would be that DVB-T2 would be deployable in by the end of 2010 at the earliest. By deployable we denote the existence of both transmission equipment and consumer reception equipment. Transmission equipment tends to lead consumer equipment by a year. Making HDTV dependent on DVB-T2 could delay HDTV unnecessarily. Ofcom has been optimistic in the bit rate improvement of DVB-T2. The objective behind the standardisation work is to achieve at least a 30% improvement with minimal reduction in coverage. Ofcom has made no reference to coverage reduction. The extent of improvement and the effect on coverage is not possible to know at this stage, but does imply a degree of uncertainty. We understand that Ofcom assumes that just three HDTV channels would be launched in the HD-multiplex in 2009, rising to four in The bit rates may not be consistent with the BBC Executive s aspirations for HDTV in terms of quality at least initially. That said, it might be argued that the quality is at least sufficient to get HDTV launched. Introducing DVB-T2 could potentially create compatibility issues with receivers on rescans. If a receiver finds a DVB-T2 channel that it cannot properly decode, it should skip over it. This is the response of receivers to analogue channels at present. However, this response is untested with DVB-T2 and a few existing 20

28 receivers may crash or hang if they receive a DVB-T2 Multiplex but are unable to decode it. The scale of this potential problem cannot be assessed for certain at this stage in the absence of DVB-T2 signals but it is not likely to be very significant. If necessary, some equipment may be upgradeable through over-air downloads. It would nonetheless be advisable to conduct receiver tests once DVB-T2 signals are available. Tolerance to DVB-T2 should be added into the D-Book specification for receivers as soon as is practicable. 4.3 Conclusions There are many uncertainties in predicting capacity. The best case would probably be to convert all six multiplexes to DVB-T2 which would give a total of about 190 Mbps. If all output were converted to MPEG-4, we estimate that there could be space for around 30 HDTV channels in In practice there are radio channels and interactive services, so not all 30 would ever be available. However this gives an indication of the possible end point given the technologies that are expected in the next ten years. 21

29 5 The move towards HDTV 5.1 Will television become high definition? To answer this question it is necessary to consider the chain from production, through transmission and reception, to the display screen, as shown in the simplified diagram in Figure 3. Broadcaster Viewer Production Encoding Transmission Reception Decoding Display Figure 3. Schematic diagram of the production chain At present, the new production standard is already HDTV. This is mainly because programme producers seek to be able to sell their material into the USA the largest single market which now requires HDTV. Many distributors now only commission HDTV content. At the other end of the chain, there is an increasing installed base of HDTV displays, driven mainly by consumers rapid take-up of flat panel TVs. The question is whether the middle parts of the chain become converted to HDTV. This is a question on which there are many opinions. Our analysis is as follows. There will be an undoubted pressure towards HDTV television because: a Production will be in HDTV b Displays are increasingly HDTV c Other media will carry HDTV (eg HD DVDs, games, camcorders, Blu-Ray etc) d Premium content satellite and cable channels will be in HDTV Against this backdrop, terrestrial broadcasters will either face the obsolescence or downgrading of their DTT services or will adapt some of their output to HDTV. Candidate programmes would be sport, films and documentaries Pressure on spectrum will mean that the trend towards HDTV will be partially offset by the need to maintain channel choice. Note that regardless of how much MPEG 4 improves, there will always be this trade-off because HDTV requires four times the information as SDTV. Realistically therefore, it is unlikely that all TV channels will be provided in high bit rate HDTV. Some channels will be provided in SDTV quality, though this may technically be low bit rate HDTV. 22

30 5.2 How can HDTV be provided? DTT According to the latest data (Ofcom Q ) 8.38m households have DTT (Freeview). HDTV could potentially be broadcast on DTT but spectrum capacity is limited. There would be a cost to upgrading the transmitter infrastructure and there would be legacy issues in relation to consumer equipment just as there has been with digital switchover Satellite According to the latest data (Ofcom Q1 2007) 8.0m households have pay satellite and 0.89m have free-to-view satellite. HDTV can more easily be broadcast by satellite where there are fewer spectrum capacity constraints. The proposed introduction of a PSB broadcasters freesat could make satellite a more popular choice Cable According to the latest data (Ofcom Q1 2007) 3.1m households have digital cable. HDTV can also be distributed by cable IPTV and DSL According to the latest data (Ofcom Q1 2007) 43,000 households have TV over DSL. HDTV could potentially be distributed by DSL if the length of cable between the exchange and the customer s modem is short enough. (The data rates achievable using DSL depends on distance.) DSL has the lowest share currently of any of the TV distribution technologies and the number of households within range would be very small. 5.3 Does DTT need to become high definition? In view of the different ways in which HDTV could be provided, and the significant capacity and cost constraints in providing it on DTT, it is worth asking whether it needs to be broadcast in DTT at all. A key requirement for the BBC is to secure universality of its services, i.e. their availability to the whole UK population. The availability of a delivery platform is determined partly by technical factors such as where transmitters are located and partly by human and legal factors such as whether an aerial is allowed to be put up in a particular location. In the case of satellite, the footprint of the satellite signal is virtually 100% of the UK. This is reduced by two factors: Some homes are in the shadow of a tree or building and there is no location on the building where a dish can legally be installed so as to see the satellite, Planning and legal restrictions may prevent a dish being installed. Satellite dishes are regarded in planning regulations as development whereas UHF aerials are not. Dishes cannot usually be installed in conservation areas. In situations where UHF antennas cannot be placed outside, they can at least be located in the roof space. Satellite dishes cannot

31 It is generally believed that satellite is potentially available to 95% of homes once these factors are taken into account. Planning restrictions are most severe in urban areas, but these are more likely to have the alternative option of cable services. In the case of DTT, the footprint of terrestrial signals currently cover around 95% of homes. The homes not covered are either in the shadow of hills or very massive buildings, or simply too distant from the nearest transmitter or repeater. By the time switchover is complete, DTT will have effectively the same footprint as satellite. The coverage maps are not identical, so in practice terrestrial can act as an in-fill for satellite and vice-versa. However while satellite is excellent for providing fixed reception to homes, portable receivers cannot use satellite as the aerial required is too big and needs accurate alignment with the satellite. For portable reception terrestrial transmission is the only currently viable option. If we only consider fixed TV sets in the home, delivering HDTV via satellite could achieve a similar level of universality as DTT. If portable reception is required to be in HD, DTT is undoubtedly more universal. However, it may be questioned whether portable reception needs to be in HD. Here, opinions differ. If portable were not to require HD, then one scenario would be for HDTV to be provided by satellite for fixed sets and SD to be provided by DTT for portable sets. This would be a rational and economically efficient partitioning of spectrum. Since a new receiver is required for HD anyway forcing this receiver to be converted to be a satellite receiver would not be detrimental to existing terrestrial users. All the above are technological comments on the way in which different platforms might in future be used. Spectrum efficiency is not the whole story, however. Other countries such as the US, Japan, Spain, Australia, France and Italy have all made commitments to providing HDTV on the terrestrial platform. So if the UK were only to provide it on satellite it would be at odds with the majority of countries. The UK currently has a strong consumer electronics industry and the trade association Intellect has expressed fears that the UK could lose out commercially if it were not able to continue to develop and sell DTT equipment into its home market. Given that one of the rationales for digital switchover was to retain and enhance the UK s lead in TV technology, it would be ironic to erode that lead in a transition to HD. Satellite reception equipment has traditionally been supplied through a different channel and according to a different business model to terrestrial equipment. If HD were to be distributed via a non-subscription satellite service, the supply chain would need to adapt, and different business models would need to emerge. The BBC and other PSB broadcasters are intending to launching a freesat service which could, if successful, deliver satellite equipment through the same distribution channels as DTT receivers. Like all change, there could be both opportunities and threats for the supply chain. Though satellite is potentially an option for most households, some consumers may be resistant to changing from an aerial to a satellite dish. They may be uncomfortable with using new supply channels. There are many unknowns here, but it is at least possible that there could be an impact on universality. DTT has the advantage of being familiar to most people. 24

32 5.4 The need for a national strategy Given the arguments raised in this chapter we believe that the BBC Executive s proposed service should be situated in the context of a coherent national strategy regarding HD. It is clear that there is a momentum towards HD but the importance and appropriateness of broadcasting HD on the DTT platform is not self evident. It needs to be assessed carefully taking into account the wider interests of UK plc such as broadcasters and equipment manufacturers. It is not clear that there has yet been a forum in which this has been done. With a national strategy it will be possible to make proper decisions for the supply chain: for example, whether to plan to change receivers to MPEG-4 anddvb-t2. The BBC has led broadcasting technology over many decades, participating in standards setting and international bodies. This lead is assumed to help broadcasters and equipment manufacturers, and to give rise to an economic benefit. In the case of HD there is still technical and standardisation work to be done, and the BBC has a chance to take a leadership role. 25

33 6 Switching DTT to high definition the broadcasting perspective 6.1 The scope for introducing one BBC HDTV channel in the BBC s currently envisaged allocation A net increase of 9 Mbps across the BBC multiplexes would be sufficient to carry a single basic HDTV service in 2012 but in our view such a service would not meet the high quality requirements historically adopted by the BBC and endorsed by the BBC Executive. However, it may be that the bit rate required for the existing SDTV MPEG-2 channels will decrease. While the performance of MPEG-2 has tended to plateau, some improvement may still occur. Alternatively a decision to sacrifice picture quality on the SDTV channels might be made. If a 20% reduction in bit rate for current picture quality can be achieved (less improvement than has been achieved over the last 5 years) approximately 8 Mbps further capacity might be made available. Together with the 9 Mbps gain from mode change, and other changes, a total of 17 Mbps could become available which would be sufficient for one high quality HDTV stream. The 17 Mbps would not all be released on one multiplex, however. To get enough capacity for an HDTV channel on one multiplex would require a reallocation of channels by the BBC and also those third party broadcasters channels which are also planned to be carried on the BBC's two multiplexes. This might involve moving some third party channels from the BBC multiplexes to elsewhere - a development which would require co-operation across all PSB broadcasters. If initiated by the BBC alone there would be few incentives for the other PSB broadcasters to assist in the ways indicated. Consequently we believe that this is a possible but unlikely option. 6.2 One-off events in HDTV If there was insufficient capacity to offer a continuous HDTV channel on DTT, it might be feasible the BBC to broadcast one-off events in HDTV. The BBC already adapts the capacity available to support different TV channels (e.g. CBBC and BBC 3 share the same capacity). To achieve this in practice it might be necessary to rearrange the channels on the multiplexes so that all the channels that could conveniently be suspended for an HDTV transmission are on the same multiplex. The number of channels that would have to be taken down will change over time in line with the expected improvements in MPEG-4 encoding as shown in Table 8. This table assumes that MPEG-2 channels are typically Mbps. 26

34 Table 8: Predicted ratio of SDTV channels to an HDTV channel Year Ratio : : :1 As explained above, it may be necessary to move channels over to a different multiplex to obtain enough capacity on one multiplex for an HDTV channel. The problems are different according to whether the channel moves are one-off or dynamic. In all cases there may be problems with the behaviour of recording devices being wrong footed by a channel change. Dynamic channel changes are possible in principle with newer equipment. However in practice, older receiving equipment may malfunction when trying to do this. 6.3 Accommodating five PSB HDTV channels The bit rate required per channel in 2012 will be between 6.5 and 11.5 Mbps depending on the quality sought. Assuming an average of 8 Mbps per channel (the BBC would probably want more than 8 Mbps), to transmit the five PSB channels in HDTV would require a minimum of 40 Mbps. There would also be a need to simulcast the same five channels in SDTV MPEG-2 as existing receivers will not be able to receive the HDTV MPEG-4 signal. Given the current average bit rate for SDTV MPEG-2 transmission of 3 Mbps this would require a further 15 Mbps. We estimate that the five PSB channels would therefore consume a total of 55 Mbps initially. While there is adequate capacity on the 3 PSB multiplexes to carry the five PSB channels in HDTV and simulcast, there are already other channels being carried on the PSB multiplexes as shown in Table 9. Table 9: Planned channel complement of the PSB multiplexes immediately postswitchover Existing analogue (PSB) channels (BBC1, 2, ITV1, Channel 4, Five/S4C) BBC digital PSB channels (eg News 24, BBC3 & 4, CBBC, CBeebies, Parliament) Other channels carried on PSB multiplexes (ITV2,3 & 4, More 4, E4, Film 4 +1) It would be necessary to find a new home on other multiplexes for the channels that are cleared away to make room for HDTV. With a gross capacity increase at switchover from the mode change across all multiplexes of 24 Mbps there is not 27

35 enough capacity being gained to achieve this without some channels being removed altogether. Even Ofcom s optimistic estimate of a 33 Mbps gain from switchover is inadequate for 5x8 Mbps channels. 6.4 A longer term view In the longer term, to make a substantial transition to HDTV on DTT will require more capacity to be squeezed from the allocated DTT spectrum. If all six multiplexes were converted to DVB-T2, the total capacity would rise from 144 Mbps to around 190 Mbps. If all output were converted to MPEG-4, by 2017 there could be space for around 30 HDTV channels. However this will entail the conversion of all receivers in three important respects: (i) from SD to HD, (ii) from MPEG-2 to MPEG-4, and (iii) from DVB-T to DVB-T2. 28

36 7 Switching DTT to high definition the consumer equipment perspective 7.1 The importance of converting consumer equipment to MPEG-4 If HDTV is to be introduced into DTT in a substantial way (i.e. several channels), it will be desirable to convert all DTT broadcasting to MPEG-4: Until MPEG-2 receivers are phased out, MPEG-2 will have to be simulcast with MPEG-4 HD services. On the PSB channels alone, this consumes 15 Mbps. MPEG-4 reduces the bit rate required for an SDTV channel so that a switch to MPEG-4 for these services will free up extra capacity which could be used for HDTV. We estimate that there would be an eventual doubling of capacity from switching from MPEG-2 to MPEG-4 for the same picture quality see Table 8. Existing MPEG-2 consumer equipment is not able to receive MPEG-4 7. Continuing to supply MPEG-2 equipment will lead to a situation where there remains a large installed base of MPEG-2 receivers and therefore a requirement to simulcast MPEG-2 alongside any new HDTV services. HDTV could then only be introduced after MPEG-2 transmission has ceased and made more capacity available. We believe it unlikely that MPEG-2 could be phased out in less than 10 years from the last sales of MPEG-2 equipment. If DTT receivers on sale in the UK were to all be MPEG-4 compatible by 2010, this would give the earliest date of 2020 for MPEG-4 as the norm. At this point, all TV could become just TV, some channels or programmes may be provided at low bit rate HDTV, others at high bit rate HDTV, some as SDTV. The key is that they will all be MPEG-4 and the receivers will be able to receive any of the transmissions and display them appropriately. There is therefore a task in converting consumer equipment, which is discussed next. There are two aspects of this conversion, namely encouraging the uptake of MPEG-4 equipment and dealing with the legacy of MPEG-2 equipment. 7.2 Encouraging the early uptake of MPEG-4 equipment Conversion to MPEG-4 will entail the complete replacement of all DTT receiving equipment currently installed and being installed. In view of this it would be desirable for the supply chain to stop supplying MPEG-2-only decoders as soon as possible. This section considers scenarios for encouraging MPEG-4. 7 However in the more distant future upgradeable software only decoders might appear which could be upgraded from MPEG-2 to MPEG-4 in the field. Take for example PC decoders for DTT that exist now. These have a hardware DVB-T receiver demodulator and use a software MPEG decoder. When the BBC launched the HD trial, MPEG-4 decoder upgrades for these appeared on the internet allowing many people who were not selected to receive STBs to watch the trial HD transmissions on their PCs. 29

37 7.2.1 Equipment-led scenario In the absence of MPEG-4-only services or HDTV, consumers will be indifferent to whether their decoder is MPEG-4 compatible or not. Changing to MPEG-4 in this scenario will need to be effected through changes to equipment standards, by agreement with manufacturers and retailers. An example of this is incorporation of DVB-T 8k mode receivers. The original specification and current transmission in the UK uses 2k mode. At switchover we will be changing to 8k mode. In preparation for this the current equipment specification (DTG D book) requires the support of 8k mode. The established mechanisms for influencing equipment specifications include: Incorporation of a requirement to be able to decode MPEG-4 in the D-Book. The D-Book is published by the Digital Television Group (DTG) and sets out the agreed technical standards for digital TV receivers. Incorporation of a requirement to be able to decode MPEG-4 in the digital switchover Certification Mark ( digital tick ) criteria. The 'Digital Tick logo was developed as a certification mark to help consumers through digital switchover. It certifies products and services that are designed to work before, during and after switchover. The logo is owned by the Secretary of State for Trade and Industry, with Digital UK licensed to administer and coordinate key aspects. Products are awarded the digital tick if their technical features and performance meet agreed criteria. The criteria are agreed between Digital UK, government and the supply chain. A commitment by the key parties such as the BBC, DTI/DCMS and Ofcom to make the transition. Because MPEG-4 is being adopted by other countries in Europe, the supply chain is starting to include MPEG-4 capabilities in the chipsets that are used in UK equipment. UK equipment may not include the relevant software however. Ultimately there will be little or no extra cost of including MPEG-4 in UK equipment. In the short term there will be a cost, and this will adversely affect the costs of the targeted help scheme Broadcaster-led scenario New broadcast standards are generally driven by broadcasters starting to transmit desirable services using them, which causes receiving equipment to become available. In the UK it has generally been the BBC that transmits the first services so driving equipment availability. An example of this is DAB which was launched in the 1990s into a market with no receivers. Another example is teletext, launched by the BBC and ITV in the early 1970s. Early receivers were very expensive but slowly the price came down, led by Philips investment in semiconductors for receivers. If the BBC announced its intention to switch some of its services to MPEG-4, or if HDTV were to become available, consumers would start to demand MPEG-4 receivers and the supply chain would follow. This approach is being adopted in Australia. Australia is proposing to switch to MPEG 4 by requiring new TV services (both SD and HD) to be broadcast in MPEG-4 only, so driving the take-up of MPEG 4 with the new services. 30

38 7.3 Dealing with the legacy of MPEG-2 equipment The existence of legacy equipment will be the key determinant of the timing of any switch-off of MPEG-2. The options here are as follows Non-intervention Broadcasters could simply stop transmitting MPEG-2 and leave it to consumers to change their equipment. There would need to be a small enough installed base of MPEG-2 equipment that either government, Ofcom or broadcasters are prepared to take political (but not financial) responsibility for such a change. Such a policy may seem harsh. But there is an important issue here that needs to be addressed. In some areas of consumer spending, everyone accepts that equipment has a limited lifetime. Personal computers, for example, become effectively obsolete in ten years because they can no longer run the latest software. In the past, television technology has changed sufficiently slowly that people assume that it should go on working for years. If the UK is not to face an endless series of managed switchovers, people must be educated to understand that television technology is going to behave similarly to personal computer technology and therefore a ten-year old piece of TV reception equipment should be accepted as obsolete Managed switchover The alternative approach is where the switchover is managed and paid for by broadcasters or some other public body that could pay for the replacement of old MPEG-2 boxes. This, in effect, is what has occurred with digital switchover through the help scheme. Such managed schemes are not unusual in subscription services 8 but the economics do not make it a desirable choice for non-commercial broadcasting. 7.4 The interaction of MPEG-4 and DVB-T2 A complicating factor here is that in the case of TV there are two logically separate areas of progress compression standards (eg MPEG-4) and modulation standards (eg DVB-T2). Unfortunately the two domains are not synchronised in their respective standardisation processes and MPEG-4 is several years ahead in the standardisation process than DVB-T2. This could lead to consumers changing from MPEG-2 to MPEG-4 only to find that three or four years later they need to change from DVB-T to DVB-T2. The Ofcom proposal has the benefit that it pulls these two transitions into synchrony. 8 Where there is a subscription service and associated equipment there is a strong incentive for the service provider to manage equipment replacement. In the mobile phone industry there has been continuous change over the last 20+ years. Initially Analogue handsets from 1984, followed by more secure analogue handsets in early 1990s, then 2nd generation (digital) handsets starting in 1994, replaced by GPRS handsets supporting data services from 2001, now 3G high speed data/video handsets from There are 2 further cycles of handset replacement already in the pipeline. These continuous replacements have been driven by the operators offering subsidised handset upgrades funded on the expectation of higher revenues from the new services and lower infrastructure costs for the old services on the new platforms. A similar situation occurred with the Sky switchover to digital. Initially launched as a new platform the prices dropped as subsidies increased on the expectation of higher revenues and lower costs once the analogue broadcasts were switched off. Sky managed to switchover in 3 years using this approach. 31

39 32

40 Appendix A: Terms of reference for advice to BBC Trust on spectrum usage The BBC Trust is conducting a Public Value Test on the BBC Executive s proposal to launch a high definition television channel. The Trust wishes to make an assessment of spectrum capacity issues, independently of estimates available from both the BBC Executive and Ofcom. 1. What compression rates are currently advocated or used for HD transmission in MPEG-4 on DTT, both domestically and in the countries leading HDTV? What are the issues and trade-offs with different compression rates? Do compression considerations vary by platform (DTT, satellite, cable)? 2. How is compression technology likely to evolve? What are the implications for the use of spectrum for HDTV? By what timescale and to what bitrate might compression rates for HDTV reduce? 3. What capacity will there be, if any, after digital switchover for the provision of HDTV channels in the allocation of spectrum currently envisaged for PSBs? What additional capacity would be needed to accommodate five PSB HD channels? What would be the scope for introducing one BBC HD channel in the BBC s currently envisaged allocation? 4. If HD is the new production standard, how likely is it that at a future point all TV channels will provided in HD? By when might this occur? What are the alternative scenarios? 5. If there was insufficient capacity to offer a continuous HD channel on DTT, how feasible would it be from a spectrum capacity perspective for the BBC to broadcast one-off events in HD? How many channels would have to be taken down for this purpose, and what multiplex management issues might this create? 6. Is MPEG-4 likely to replace MPEG-2 as the codec standard for DTT? What factors might encourage MPEG-4 adoption? 7. What would be the capacity gain from a wholesale switchover to MPEG-4 on DTT? What would be the tipping point in MPEG-4 adoption that might enable this to happen? What might be a credible timescale? 8. What might be the longevity of MPEG-4 as a codec standard? Are any alternatives currently foreseen? 9. What other factors might improve spectrum efficiency? 10. To what extent would carriage of HD on DTT secure universality of HD provision? 33

41 Sagentia Many minds make bright work We create value from technology. We do this by developing richer solutions with our clients that enable them to realise better business opportunities. We operate in six market sectors developing new technologies, products and services that change the basis of competition. We assist business leaders and policy makers to create strategies for technology, innovation and growth. Our Collective Technology Wisdom the unique characteristic of our company guides how we work. We form highly creative teams that draw on individuality and collective experience. And we take a multi-dimensional approach to opportunity discovery and problem solving, drawing on our combined technical expertise, business acumen and industry experience. We can work with you wherever you are in the world. Our teams are situated in state-of-the-art facilities in Cambridge UK, Frankfurt and Stockholm in Europe, Boston and Baltimore in the USA, and Hong Kong and Shanghai in China. info@sagentia.com Sagentia Ltd Sagentia GmbH Sagentia Inc Sagentia Inc Harston Mill Westend Carree Reservoir Place Cronhill Drive Harston Grüneburgweg Trapelo Road Suite B Cambridge D Frankfurt Suite 154 Owings Mills CB22 7GG Germany Waltham MA Baltimore MD UK USA USA Sagentia Catella AB Veddestavägen 7 SE Järfälla Sweden SGAI Tech Ltd Unit 5&6, 10/F, Block B Veristrong Industrial Centre 36 Au Pui Wan Street Fotan Hong Kong T T T T T T