Why mobile reception?

More recently, the interest arose on the use of DVB-T receivers on moving cars in order to deliver video as well as other digital data that can be allocated on the available spectrum. These new applications are a promising opportunity, and may represent a strong competitive advantage for terrestrial broadcaster, while satellite and cable DVB may be predominant for static (home) reception. In this respect, the ability of the existing DVB-T standard to ensure good quality reception for mobile applications is also a relevant issue in the comparison with other competing digital video broadcasting standards.

What's the challenge?

However, the transmission at an high bit-rate is a challenging task for DVB-T, since it employs the Orthogonal Frequency Division Multiplexing (OFDM) as modulation technique. In this modulation, data are divided into many parallel streams and each stream is modulated on a different sub-carrier frequency. The block-processing structure of OFDM yields that a proper reception is possible only if the channel is time-invariant at least for the duration of one block. If this condition is not satisfied and the channel changes during the transmission of each block, then interference arises among the different sub-carriers (inter-carrier interference, ICI).

Where do we stand?

Various solutions to counteract the effects of the interference have been proposed. Longer time interpolators are useful for a better estimate of the channel, as shown by Espineira and Stare (1), while multiple antennas at the receiver yield a diversity gain, see Faria G. (2), that overcomes partially the performance loss due to time-varying channel. However, both solutions have important drawbacks, since the interpolators require more silicon area for memory, while the use of multiple antennas yields an additional cost due the extra equipment and its deployment. On the other hand, current solutions based on one antenna are exceedingly complex for the DVB-T. Jeon W. G., Chan K. H. and Cho Y. S. (3) proposed a linear equalisation scheme, which requires operations with high complexity. An alternative, low complexity approach has been proposed by Gorokhov and Linnartz (4-5). They consider a scheme where the channel parameters and performing a cancellation of the interference, but their application of these schemes to the existing DVB-T standard is not straightforward, since a training sequence was used in order to estimate the channel parameters, which is not present in DVB-T standard. Moreover, when sub-optimal channel estimators are considered, for high constellation and high speed the schemes do not guarantee the necessary QoS.

We propose a new complexity scheme for the channel parameter estimation. Moreover, in order to increase the maximum speed that can be reached with the required QoS we consider an iterative estimation of the channel parameters and the interference cancellation. The main advantages of the proposed systems are: a higher speed at which the DVB-T reception is correct, a complexity similar to the previously-proposed two antennas system, the possibility to be used on a two antenna system, with additional benefits. Simulations results are presented for various modes, constellation sizes and code rates of the DVB-T standard. By comparing the performance of the reduced complexity schemes with existing multiple antenna solution, we conclude that the new techniques allow the use of DVB-T for similar or higher speeds range with a limited complexity.

How does it work?

See the detailed discussion

Maximum attainable speed

A first set of simulation results show the maximum speed at which the DVB-T receiver can work on the channel CH40 (626 MHz) with a bandwidth of 8 MHz. Note that the overall band for analogue TV is between 400 and 790 MHz, so that the considered channel is roughly in the middle of the spectrum. Since the effect of mobility is stronger at higher frequency, this choice gives an average behaviour of the schemes.

Figure 3 – Maximum achievable speed for various schemes.

Figure 3 shows the maximum speed that can be reached by the DVB-T system adopting different modes and constellations for the following equalization/estimation modes:

• Continuous line, black circles -- One antenna, standard equalisation and channel estimation methods (no ICI cancellation), compare Figure 8 of Faria G. (2).
• Continuous line, empty squares -- Two antennas with maximum ratio combining, standard equalisation and channel estimation methods (no ICI cancellation), compare Figure 8 of Faria G. (2).
• Dashed line, black squares -- One antenna with iterative ICI cancellation, 1 iteration, error correction, Taylor coefficients estimation in the loop.
• Dashed line, empty squares – Two antennas with ICI cancellation, no error correction, Taylor coefficients estimation by training sequence. Using a training sequence the channel estimator is performing better.

From Figure 3 we see that the ICI cancellation scheme performs always better than the two antennas system. Most of current interest is focused on the 8k mode, since it allows an efficient deployment of Single Frequency Networks. On the other hand, higher constellation sizes as well as lower code rates are preferred since they yields a higher bit rate. Hence, on Figure 3 the attention should be focused on the 8k / 64-QAM results. Even if the performance of one-antenna system with ICI cancellation are much better than no-ICI-cancellation schemes, still for the interesting 2/3 code rate the maximum speed falls below the 100 km/h. Only by using the combination of ICI cancellation and two antenna higher speeds are reachable.

In the table we also included the results with two antennas and advanced channel estimation, as obtained by Espineira and Stare. In this case two antenna are used and no ICI cancellation is considered. The signals from the two antennas are MRC combined before decoding. Time interpolation (12 pilots), frequency interpolation (16 pilots).

Table 1 – Maximum speed range for 64-QAM @ 8k.

NEXT