Single-frequency Network - DVB-T SFN

DVB-T SFN

In DVB-T a SFN functionality is described as a system in the implementation guide. It allows for re-transmitters, gap-filler transmitters (essentially a low-power synchronous transmitter) and use of SFN between main transmitter towers.

The DVB-T SFN uses the fact that the guard interval of the COFDM signal allows for various length of path echoes to occur is not different from that of multiple transmitters transmitting the same signal onto the same frequency. The critical parameters is that it needs to occur about in the same time and at the same frequency. The versatility of time-transfer systems such as GPS receivers (here assumed to provide PPS and 10 MHz signals) as well as other similar systems allows for phase and frequency coordination among the transmitters. The guard interval allows for a timing budget, of which several microseconds may be allocated to time errors of the time-transfer system used. A GPS receiver worst case scenario is able to provide +/- 1 us time, well within the system needs of DVB-T SFN in typical configuration.

In order to achieve the same transmission time on all transmitters, the transmission delay in the network providing the transport to the transmitters needs to be considered. Since the delay from the originating site to the transmitter varies, a system is needed to add delay on the output side such that the signal reaches the transmitters at the same time. This is achieved by the use of a special information inserted into the data stream called the Mega-frame Initialization Packet (MIP) which is inserted using a special marker in the MPEG-2 Transport Stream forming a mega-frame. The MIP is time-stamped in the SFN adapter, as measured relative the PPS signal and counted in 100 ns steps (period time of 10 MHz) with the maximum delay (programmed into the SFN adapter) alongside. The SYNC adapter measures the MIP packet against its local variant of PPS using the 10 MHz to measure the actual network delay and then withholding the packets until the maximum delay is achieved. The details is to be found in and mega-frame details in.

It should be understood that the resolution of the mega-frame format is being in steps of 100 ns, whereas the accuracy needs can be in the range of 1-5 us. The resolution is sufficient for the needed accuracy. There is no strict need for an accuracy limit as this is a network planning aspect, in which the guard-interval is being separated into system time error and path time-error. A 100 ns step represents a 30 m difference, while 1 us represents a 300 m difference. These distances needs to be compared with the worst case distance between transmitter towers and reflections. Also, the time accuracy relates to near-by towers in a SFN domain, since a receiver is not expected to see the signal from transmission towers being geographically far apart, so there is no accuracy requirements between these towers.

So called GPS-free solutions exists, which essentially replaces GPS as the timing distribution system. Such system may provide benefit in integration with transmission system for the MPEG-2 Transport Stream. It does not changes any other aspect of the SFN system as the basic requirements can be met.