Mobile reception on trial
By Mike Brooks
The arrival of digital television has opened up the possibility of receiving high-quality and steady pictures in a moving vehicle. Here, NTL Broadcast trials mobile digital television in the UK to investigate the parameters that make a good mobile receiver.
Although the word unique is overused, it can truly be applied to the ability of the terrestrial TV platform to provide a service on-the-move. While analog technology was too prone to fading and delayed signals to be of practical value, the arrival of digital TV has opened up the possibility of receiving high-quality and steady pictures in a moving vehicle. A high-bandwidth delivery pipe of this nature has applications beyond television. Broadcasting high-speed digital information specifically to mobile receivers is an untapped market with considerable potential and worldwide interest. Whatever the use and business implications, however, the technological aspects of mobile reception need to be fully understood.
Digital terrestrial television (DTT) was introduced first in the United Kingdom, where a standard-definition multichannel service began in November 1998. NTL Broadcast built the complex networks for Digital 3 and 4 -- the company established to handle the digital output of national commercial stations ITV and Channel 4 -- and for SDN. The networks include video compression and multiplexing centers in addition to the transmitters themselves. No consideration was given to mobile reception, as the priority was the early introduction of a digital terrestrial platform for fixed reception on ordinary rooftop antennas, with as many channels as possible.
The Digital Video Broadcasting terrestrial format (DVB-T) was adopted by the UK, with MPEG-2 video compression and COFDM using approximately 2000 carriers and 64 QAM modulation. This provides about 24 Mbits/s of data per UHF radio-frequency channel, which is sufficient for about six TV services. In most areas of the UK, six multiplexes can be received, giving a total capacity of about 30 to 40 channels, but because no new spectrum was made available, the digital channels had to be found from spare frequencies that were taboo for analog TV. For mobile reception, a more rugged form of transmission, using plain (but extravagant) QPSK or perhaps 16 QAM modulation, was considered necessary.
Birth of mobile
Having gained experience in building DTT networks in the UK, NTL Broadcast won a contract to build a transmitter network in Singapore specifically for mobile reception. Opening in early 2001, this was the first commercial application of mobile DTT in the world. It provides a TV service to buses throughout the island using a single-frequency network (SFN) of 10 transmitters.
Experience with this network showed that in a mobile environment, the performance of the receiver is a particularly critical element in the success of the whole end-to-end system, but because NTL was not responsible in Singapore for receivers, it decided to mount a UK trial comprising laboratory as well as field work in order to investigate the parameters that make a good (or bad) mobile receiver. It was thought the trial would also be useful in confirming transmission power and data rate issues, and the balance between the two.
Two transmitter sites were set up in the Oxford area, using a single frequency and carrying two TV services. One contained news material with subtitles, and therefore required no audio, and the other was the output of a local TV station, which did carry audio. This provided realistic content that helped in the practical assessment of TV material in a mobile environment-- in this case, a test vehicle fitted with seat-back LCD screens.
Even with digital technology, mobile reception is a challenge for three main reasons. First, Rayleigh channel characteristics mean that with no guaranteed direct line of sight to the transmitter, multipath reception becomes the norm. Second, Doppler shifting of echoes due to motion of the vehicle causes Fast Fourier Transform (FFT) leakage through carrier spreading and, hence, a loss of orthogonality between carriers. Third, rapid channel state changing due to vehicle motion makes channel estimation, i.e. compensation for amplitude and phase variations across the multi-carrier block of frequencies, difficult. Furthermore, in an SFN, reception situations arise that can confuse the unsuspecting receiver. For example, the closest transmitter may not be the strongest while it is blocked by a hill or building. The more distant transmitter then becomes the reference source for a time, meaning that the earlier weaker signal may cause inter-symbol interference. These difficulties are in addition to lower antenna height and gain as compared to a fixed rooftop installation.
It emerged that there are significant differences in receiver performance due to the chosen COFDM demodulator chip's ability to deal with these issues. So far, there seems to be no perfect design. Some are good in a rural (long echo) environment, while others perform best in urban (short echo) conditions. Some work well with a single transmitter, but are poor in an SFN. In certain environments, some perform almost as well in 16 QAM as in QPSK mode. Having the entire system under test conditions made it possible to vary parameters at will, and allowed valuable performance data to be collected that would help in the optimum design of a mobile receiver.
Possibilities and options
The Oxford trial confirmed that consistent high-quality video, audio and data can be delivered to moving vehicles using existing frequency spectrum. It also confirmed that, for a given receiver, lower data rates delivered more robust signals. Experiments showed however, that with careful choice of demodulator chip, reliable reception could be maintained at higher data rates with increased transmitter power. Within the DVB-T specifications, a broadcaster can choose one of three modulation constellations, namely QPSK, 16 QAM and 64 QAM, and then select one of five FEC ratios. Selecting QPSK modulation with 2/3 FEC ratio would allow a data rate of up to 8.04 Mbits/s.
Selecting 64 QAM at the same FEC ratio would allow a data rate up to 24.13 Mbits/s, but would require a fifteenfold, or approximately 12 dB, increase in signal strength for acceptable reception. This allows a broadcaster to tailor its services to the target population. Because the number of services within a multiplex depends on available data rate, and the area covered by a transmitter depends on the robustness of the signal and its power, the parameters chosen for a DVB-T transmission become commercially driven.
In well-populated countries, it may be difficult to find a single frequency that is available for a national mobile service, such as might be received on intercity buses or trains. Spectrum planners, however, may well be able to find several frequencies for use in sub-national wide-area networks, with seamless transition between areas. Coverage could be concentrated along main routes, rather than centers of population, in order to bring the benefits of broadband multimedia communications to travelers and commuters.
Furthermore, using cellular telephone networks for the return path would allow interactivity and extract the maximum benefit from the broadcast data, thus helping to meet the rising expectations of the mobile public.
Digital terrestrial television has the potential to offer perfect pictures and other data on the move. This unique feature allows applications that could increase the scope of the digital terrestrial platform for broadcasters and viewers alike. However, a practical understanding of the issues involved is essential so that when important decisions are made about future services, they are based on reality and not theory.
Mike Brooks is technology manager at NTL Broadcast.
Do you have a comment about this article? To tell us your thoughts, click here.