ETSI
TSI TR 101 190 V1.2.1 (2004
1)
3
A longer guard interval could compensate longer echoes:
• lengthening the guard interval without changing the absolute duration of the useful interval would accordingly
decrease the channel capacity, thus reducing the deliverable bitrate;
• alternatively, lengthening both the guard interval and the useful interval would not bring any penalty to the
channel capacity, but would make the signal processing more difficult because of the higher number of carriers
that would result from the larger symbol duration.
Table 1 summarizes the possible lengths of the guard interval specified in the DVB-T (see EN 300 744 [5])
specification depending on the chosen FFT length.
Table 1: Specified lengths of the guard interval
Length of the guard interval Proportion to the
length of the useful interval
8k-mode 2k-mode
1/4 224 µs 56 µs
1/8 112 µs 28 µs
1/16 56 µs 14 µs
1/32 28 µs 7 µs
The longer guard intervals are suitable for networks with longer distances between the particular transmitter station, as
for example with national single frequency networks. The shorter intervals are suitable for regional or local broadcast
transmissions.
According to table 1, there are two different modes regarding to the number of carriers. The length of the useful
interval is 896 µs for the 8k-mode and 224 µs for the 2k-mode. Due to the orthogonality of the system, this corresponds
to a carrier distance of 1 116 Hz and 4 464 Hz, respectively.
One basic requirement for the DVB-T system was the bandwidth constraint in order to match an 8 MHz channel
spacing. From this requirement one can derive the number of possible carriers. 6 817 carriers per OFDM symbol for the
8k-mode (6 048 useful, the others for synchronization and signalling) and 1 705 carriers per OFDM symbol for the
2k-mode (1 512 useful carriers) are specified in the DVB-T system. The OFDM symbols can be calculated by the
Inverse Discrete Fourier Transform (IDFT). Virtual carriers are inserted in such a way that the total number of carriers
becomes a power of two, so that the faster algorithm of the Inverse Fast Fourier Transform (IFFT) can be used. At the
receiving side, the corresponding signals can be easily recovered using the respective 2k-FFT or 8k-FFT.
In order to ensure robust transmission of the OFDM signal, an error protection code is applied. In addition to the fixed
algorithm of energy dispersal, block coding, outer and inner interleaving, a Rate Compatible Punctured
Convolutional (RCPC) code has been defined as in the DVB Satellite standard. The mother code has a constraint length
of 7 bits and works with a code rate of 1/2. The two generator polynomials of the convolutional encoder are 171 and
133 in octal notation.
To adapt the error protection to the actual transmitting conditions, several code rates can be chosen. The following code
rates are specified in the DVB-T (see EN 300 744 [5]) (and DVB Satellite (DVB-S) (see EN 300 421 [3])) system:
1/2, 2/3, 3/4, 5/6, 7/8
The code rate 1/2 has the highest redundancy, but the highest transmission safety. This mode should be applied to
strongly disturbed channels. On the other hand a code rate of 7/8 has a low redundancy but a very weak error protection.
Therefore, it should be used for channels with only low interference.
As mentioned above, every carrier is modulated by a modulation symbol. QPSK, 16-QAM and 64-QAM are used as
modulation methods, e.g. 2, 4 or 6 bits per modulation symbol. The bits are assigned to the particular points in the phase
space according to the so called Gray-code mapping. The advantage of this mapping is the fact that closest constellation
points differ only in one bit. The constellation diagrams for each modulation method are illustrated in figure 1.