1993 - Ralf Schäfer - Heinrich-Hertz-Institut für Nachrichtentechnik Berlin GmbH

This paper summarizes the different activities in the field of digital TV/HDTV broadcasting in Europe, shows the way how the different activities are coordinated and explains the relations between the German HDTV-T project and the other European projects. Afterwards some details on system aspects, on techniques used for channel coding/modulation and on source coding are presented.

In the development of Advanced TV systems very rapid and dramatic changes are actually taking place. The whole development is no more driven by HDTV but by digital TV over satellite and cable. This is no longer only true for the U.S.A. with its big cable projects like those of Quantum or of Tele-Communications, Inc. (TCI) with more than 150 digital TV channels and the satellite project Direct-TV with more than 100 digital TV channels, but this is now also true for Europe.

The Luxembourgean satellite operator SES has announced, that after the launching of their two satellites ASTRA Id and l.e in 1995, a system capable to transmit 180 digital TV channels will be operational. Therefore the different European research activities in the field of digital TV and HDTV over terrestrial channels (dTTb, STERNE, SPECTRE, HD-DIVINE, HDTV-T) and over satellite (HD-SAT, FLASH-TV) have to take into consideration these new circumstances.

In the framework of the European Launching Group (ELG) and of the Working Group on Digital TV-Broadcast (WGDTVB) efforts are made to co-ordinate all the different R&D activities and to develop a common concept for digital TV- and HDTV-broadcasting over cable, satellite and terrestrial channels Europe. The HDTV-T project represents an import part of the European research developments and the collaboration is documented by the participation in the WGDTVB and by the cooperation agreement with dTTb. It is the objective of the HDTV-T project, to develop and build up a complete transmission chain for digital terrestrial broadcast of TV and HDTV until mid 1995, in order to carry out field trials and to optimise the system.

The European Launching Group is an association of manufacturers (Philips, Thomson, Grundig, Nokia etc.), of broadcasters (ARD, ZDF, BBC, SVT etc.), of operators (Telekom, France Telecom) and of administrations (German Ministry of Post and Telecommunications, British Department of Trade and Industry etc.). This association has been initiated by the same group which is responsible for the PALplus activity and which then merged with a similar activity started by the German Ministry of Post and Telecommunications. It has to be understood that this activity has been created on a voluntary basis and its participants originate from the high level managements of the member organisations. The ELG receives its strategical and technical proposals from the WGDTVB, which is the real co-ordination group in Europe, however, without a legal mandate by the European Community or by other official institutions.

The WGDTVB is composed of about the same organisations as the ELG with additions of some research institutions. These members represent the most important institutions in Europe which are involved in the development of digital TV and HDTV in Europe and they represent also the different national (STERNE, SPECTRE, VIDINET, DIAMOND, HDTV-T) and international (dTTb, VADIS, HD-DIVINE, HD-SAT) research projects.

In the beginning the WGDTB has concentrated its activities on terrestrial broadcast, however, since the beginning of 1993 satellite transmission has gained much more importance due to the reasons mentioned in the introduction. The proposal of the ELG, in which a system capable to transmit digital TV and HDTV in a compatible manner is proposed, has been worked out by the WGDTVB /1/.

The proposal of the ELG consists in a hierarchical system with three levels. The system should allow, within one terrestrial channel, simultaneous reception at HDTV quality via fixed rooftop aerials, reception at EDTV quality via fixed rooftop aerials, and reception at SDTV quality via set-top (or built-in) aerials. In the last case, SDTV reception is designed for portable receivers.

In the case that this objective may not be technically possible and a fall back needs to be defined, the target should be a two-level hierarchical system which allows simultaneously, stationary reception at HDTV quality via fixed rooftop aerials, and portable reception at SDTV quality via set-top (or built-in) aerials. In this latter case, SDTV is designed for portable receivers.

Moreover the system should be reconfigurable in order to transmit two EDTV and/or SDTV programs instead of one HDTV program.

If the three-level system mentioned above proves practical, the alternative configuration should be to carry, in the same channel, two independent EDTV services, receivable on stationary receivers via rooftop aerials, which are also simultaneously receivable in a portable environment at the SDTV level, via a built-in or set-top aerial. If it proves necessary to concentrate on the two level hierarchy mentioned above, the alternative configuration should be to allow, in the same channel, two SDTV quality level services, receivable in a portable environment.

The HDTV-T project has actually 10 members and more partners have applied for participation. Some of the partners receive funds from the German Ministry for Science and Technology, others participate on a self funding basis. The project is organized in four main working parties (AG1... AG4) and it is led by a steering committee chaired by the Heinrich-Hertz-Institut.

The work is carried out in three overlapping phases and most of the work of phase one is almost completed [2].

Any new system needs to have a consumer benefit in order to be introduced. The transition from analogue to digital transmission is not only a new technology for realization of the same service but allows for more flexibility. It will be possible to use transmission channels that cannot number of TV programs that can be transmitted at the same time. The use of strong data compression allows to transmit HDTV signals in the same channels nowadays used for analogue PAL or SECAM standard resolution programs. It will be possible to exchange picture resolution against number of programs by splitting the channel into several subchannels that each can carry one program of today's quality.

Finally it will be possible to establish a hierarchical system that gives the user the utmost flexibility to find its own cost/performance compromise (refer to figure 1): He may purchase a large screen HDTV receiver for his living room for the full HDTV quality but the appropriate price, on the other hand he may buy a 14" portable receiver for reception of the same channel but only with a TV resolution decoder which can be used without a large roof-top antenna.

Such a system needs to have an appropriate modulation technique that transmits different parts of the data with different ruggedness. The application of such techniques makes the digital system competitive even in fringe areas. Classical digital transmission systems show a failure characteristic with respect to worsening channel conditions that is quite different from analogue systems: While analogue systems show an increase in noise related to the S/N ratio found in the channel, the digital system due to the error correction in use will show a non impaired picture of unaltered quality down to the threshold in S/N where the error correction scheme can no longer cope with the transmission errors and the picture will completely break down. For the sake of service quality in fringe areas such a cut-off behaviour is not acceptable. The digital system should degrade step by step and show a graceful degradation. Graceful degradation can be achieved by two means: different error protection schemes for data of different priority [3] and multilevel modulation techniques [4].

In the same way as the picture quality is arranged in a certain hierarchy, the sound quality should allow for steps in quality: While the HDTV picture should carry along 5 channel surround sound, the first step down in service quality could result in two channel stereo sound and finally the portable receiver might work with mono as the lowest level. All sound information will be transmitted in data reduced form [5] in order not to waste too much of the scarce data capacity in the terrestrial channel.

The service planning is somewhat more complex in Europe than in the United sates. There is the need for nation-wide coverage of more than one TV chain that restricts the channel allocations quite a bit. In addition there are different channel width, 7 MHz in VHF and 8 MHz in UHF, this situation even differing from country to country. Different to the US, the cable distribution needs to be considered as well, since in Europe this is just another transmission medium and not a separate competing business. Taking all these boundary conditions into account the European research concentrates on OFDM (Orthogonal Frequency Division Multiplexing) modulation techniques [6] that allow for single frequency network (SFN) operation: One TV channel is given to one TV chain nation-wide, all energy arriving at an aerial, may it be directly or via multirjath or even from a distant transmitter that nowadays causes interference, can be constructively used in the demodulation process. However, the same technique might be used in standard service planning and will lead to a possible increase in the number of usable channels, because of the possibility to lower the transmission power and to shape the spectrum for minimum interference [7]. A possible introduction scenario could be the following: start with digital channels locally where capacity is available; as soon as the receiver population has grown far enough, the first analogue PAL channels could be replaced by full digital ones with the freedom to make a new frequency planning. This can finally result in an SFN operation for optimum spectrum utilisation.

A very important topic is the system multiplex, in order to use the available channel capacity in an optimal manner and to offer a large flexibility to the channel operator and the program providers. It is believed that the proposals being developed within MPEG 2 will fulfil the demands in this respect and therefore the MPEG multiplex syntax is under study.

As mentioned in the previous chapter, OFDM is the most probable candidate for terrestrial transmission. In OFDM the bandwidth of the transmission channel (for example: 8 MHz for UHF-TV channels) is split into a large number of subchannels, and a typical number of subcarriers is in the range of 256 - 8192. Furthermore the channel is devided into timeslots. If we take fs as the subcarrier distance and Ts as the duration of one timeslot, it has been proven [8] that the carriers can be orthogonal if we meet the condition:

The modulated subcarriers overlap in the frequency domain but can be separated perfectly if this condition is met. The spectral efficiency therefore is very high. The OFDM-process can easily be realized via inverse FFT, the demultiplexing via FFT, as it has been shown in several publications. The duration Ts of one symbol can be up to 1 ms. If we introduce a guard inverval, we can realize a completely echo-resistant system. Each block of duration Ts and fs can carry one modulated symbol. One can choose different modulating schemes like n-PSK or m-QAM for the subcarriers. In Digital Audio Broadcast (DAB) a QPSK-modulation has been selected. This is combined with a convolution^ channel coder. The total spectral efficiency including the losses due to the guard interval and synchronization symbols is about 1 bit per second per Hz. For HDTV applications this is not sufficient, therefore more complex modulations schemes will be used.

Investigations concentrate on schemes like 32- or 64-QAM to meet the requirement of squeezing 30 Mbit/s into one TV-channel. Of course these modulation schemes reduce the system robustness, however, the full HDTV-quality is only required under less critical reception conditions (reception with roof top antenna), as mentioned above.

To meet the demands concerning graceful gradation and portable reception additional measures are planned. These include non-uniform modulation to provide a robust transmission of high priority bits [4]. Furthermore unequal error protection will be introduced. Several error protection approaches based on combined coding and modulation are investigated. Block codes are developed as well as convolutional codes. The performance of these different systems have to be checked very carefully via field experiments.

Video source coding is used in order to achieve data compression of the incoming HDTV or TV signals from several hundreds of Mbit/s down to a net bit rate in the order of 20-25 Mbit/s considering the channel coding and modulation techniques described in chapter 7. Furthermore, the source coding scheme has to support the system requirements emphasized in chapter 6.

To meet the needs of a hierarchical HDTV/TV system and of graceful degradation, a scalable video source coding scheme under investigation in the current phase of MPEG is used, with two stages of HDTV and TV resolution. Due to international acceptance interlace sources are considered as a fact although being unfortunate because of putting constraints on source coding efficiency. Hence, for HDTV input a sampling raster of 1440x1152/50/2:1 with 54 MHz sampling rate is chosen; this helps in meeting the limited channel capacity which would be harder in case of 72 MHz sampling, and it eases deriving a TV portion of compatible CCIR Rec. 601 type resolution (720x576/50/2/2:1) for scalability.

A simplified source codec block diagram is shown in Fig. 2. The source encoder consists of two branches for layers, one for coding the TV portion derived from the HDTV input via filtering and subsampling ("down converter"), the other for coding the remaining information necessary to recover full HDTV. Via the link (up-converter) between the HD and TV encoders TV data are fed to the HD encoder for improved predictive coding. After buffering including rate control at least two data streams of different priority are created and multiplexed together meeting the bit rate demands of the FEC/modulation stage. From the TV stream a portable receiver may decode a signal with TV resolution whereas from both the TV and HDTV streams full HDTV is decoded by a stationary receiver. Each of these streams may further be split into substreams for different error protection or/and non-uniform modulation in order to improve graceful degradation ability.

The coding algorithm which has been selected is based on MPEG 2 and the means to achieve a hierarchical approach are based on "spatial scalability" and on "noise scalability" [9J. The block diagram of a three level scalable codec is shown in Fig. 3. The HDTV signal is down filtered and subsampled and the resulting TV-signal is encoded, locally decoded and up-converted. The upconverted signal is used to predict the HD-signal, which is encoded with the same MPEG 2 scheme as the TV signal.

If a three level approach is selected the TV signal can be split into two quality levels (SDTV+ EDTV) by using requantization and two different VLC encoders. When writing this paper two level and three level codecs were under study and decisions concerning the final system were expected for summer 1993.

Several European research and development activities have been started and the HDTV-T project can be considered as an important contribution towards digital TV and HDTV.

It is expected that digital TV services via satellites will be introduced in Europe at about 1995 and therefore it is now the main objective of the HDTV-T project to create a TV/HDTV - system for cable and terrestrial transmission, which will then be compatible with the existing digital TV infrastucture.

Therefore the system will rely on MPEG 2 source coding and MPEG 2 multiplex which will offer a large exent of flexibility. A hierarchical coding scheme will be used based on spatial and eventually on noise scalability, depending on the decision whether a two or three level system will be selected.

As far as modulation, channel coding a frequency planning are concerned much research is still necessary due to the lack of available spectrum and the different infrastructural and political conditition in the different parts of Europe. This work is carried out in close cooperation with RACE-dTTb and a common solution for Europe is expected with help of the coordination of the WGDTVB.

• [1] WGDTB 1063,"Report to the European Launching Group on the Prospects for Digital Terrestrial Television", November 1992
• [2] Schafer, R . et al,"HDTV-T - A Joint Research Project on Digital Terrestrial Broadcast of HDTV", Proceedings of International Workshop on HDTV 92, Kawasaki, Japan, November 18-20, 1992
• [3] K. Joseph, S. Ng, D. Raychaudhuri, J. Zdepski, R. Saint Girons, T. Savatier, "MPEG++: A robust compression and transport system for digital HDTV", Image Communication, Vol. 4, Nos.4-5, August 1992, pp.307-323
• [4] K.M. Uz, K. Ramchandran, M: Vetterli, "Combined multiresolution source coding and modulation for digital broadcast of HDTV", Image Communication, Vol.4, Nos.4-5, August 1992, pp.283-292
• [5]    MPEG CD on audio
• [6] R. Monnier, J.B. Rault, T. de Couasnon, "Digital television broadcasting with high spectral efficiency", Proceedings of International Broadcasting Convention, Amsterdam, 3.-7. July 1992, pp.380-384
• [7] A. Mason, G. Drury, N. Lodge, "Digital television to the home - when will it come?", Proceedings of International Broadcasting Convention, Brighton, September 1990, pp. 51-57
• [8] M. Alard, R. Lasalle, "Principles of modulation and channel coding for digital broadcasting for mobile receivers", EBU Review-Technical N.224 (August 1982)
• [9] G. Schamel: "Graceful degradation and scalability in digital coding for terrestrial transmission", Proceedings of International Workshop on HDTV '92, Kawasaki, Japan November 18-20,1992