Minimizing Delays in Single Frequency Networks

March 5, 2003
Last month I explained how a single frequency network (SFN) could improve DTV coverage by filling in shadowed areas and by providing the stronger field strengths required for indoor reception. A single frequency network, as the name implies, consists of multiple transmitters operating on the same channel carrying the same DTV datastream. However, as I explained, single frequency networks are not easy to design! If a DTV tuner receives signals from two or more transmitters, the extra signals will appear either as multipath or interference, depending on the time delay between the signals and the performance of the DTV tuner's equalizer. This month, I'll look at ways to minimize delay in an SFN and look at proposed standards for distributed transmission (DX).


The simplest SFN consists of a main DTV transmitter and a smaller on-channel booster to fill in an area blocked from the main transmitter. If reflections are ignored, it is easy to determine the time delay between the booster and the main transmitter using geometry and the speed of light. In the simplest case, where the booster is directly in the path (straight line) to the main DTV transmitter, if there were no delay in the booster, the two signals would arrive at the same time. Of course, there will be some delay in the booster. This could be significant if there is any filtering in the transmitter. Any additional processing will increase delay and make it more likely the main and boosted signals will not be treated as multipath by the DTV tuner, but rather as interference. If the population being served from the booster is confined to areas shadowed from the main transmitter, this isn't a problem. In some circumstances, however, careful selection of booster location and minimizing delay through the booster may be sufficient.

Another solution to the transmitter timing problem is to control the timing of the ATSC streams from each transmitter so they arrive at the DTV tuner with small enough delay for the undesired signal to be eliminated in the tuner's adaptive equalizer or correlated with the desired signal to improve the signal-to-noise (S/N) ratio.

Before I look at how transmitters in a Distributed Transmission Network (DXN) can be timed, let's look at the improvement that can be achieved with timing. At the IEEE Broadcast Technical Symposium, Merrill Weiss presented a table showing the increase in population able to receive the DTV transmissions from multiple transmitters with less than a specified delay. The transmitters were described my last column.

(click thumbnail)

The table above shows that 384,833 persons are predicted to have a carrier-to-interference ratio less than 20 dB at their location. For these people, the delay between the two signals is critical, as it will determine if their DTV tuner will consider the other signal as interference or multipath. Looking at the delay between the transmitters, you can see that with system timing adjustment number 2, the number of persons predicted to receive the signals with a delay less than 25 microseconds goes from 11,074 to 44,739. The number of persons predicted to receive the signals with a delay less than 100 microseconds goes from 52,106 to 109,301. The timing parameters are shown in the second table. Merrill cautioned that this is not the optimum timing and results can be improved with a more detailed analysis. The numbers and maps represent reception using an omnidirectional antenna at 30 feet. Interference will be less than that shown here if viewers use an indoor antenna about six feet above ground or an outdoor directional antenna.
Timing Adjustments
Color scheme:
Yellow <5µsec
Orange 5-10 µ
Red 10-25 µsec
Green 25-50 µsec
Lt Blue 50-100 µsec
Dk Blue >100 µsec

(click thumbnail)
Map 1: Timing Adjustment: None

(click thumbnail)
Map 2: Timing Adjustment #1

(click thumbnail)
Map 3: Timing Adjustment #3

The impact of the adjustments on interference and delay at the receive locations can be seen in the Timing Adjustments side bar.


How do we synchronize the transmitters? If we were dealing with analog TV transmitters, we would lock the carrier frequencies of the transmitters together using a GPS reference and control the timing of the video signal by using a frame synchronizer locked to the GPS reference. With DTV, it isn't as simple. The output data from the ATSC modulator is quite different from the input. The input datastream goes through data randomization, the addition of Reed-Solomon error correcting codes, byte interleaving, bit interleaving, pre-coding (to compensate for comb filtering used to reduce interference from co-channel NTSC stations), and trellis coding. Setting all of these processes to a known state to match the output datastream from another ATSC modulator is not trivial. I won't go into the details on how each of these processes is handled in a Digital Transmission System (DXS) this month but if there is interest, I may cover it in a future column.

The simple solution is to do all of this processing at one location and send that data to the distributed transmitters. If the ATSC transport stream is sent via fiber or microwave to each of the transmitters in the SFN, they will also have exactly the same signal. Timing, however, becomes complex because the delays in the fiber or microwave link have to be considered. This technique also requires 50 percent more bandwidth than the 19.39 Mbps transport stream. Studio-transmitter links running at the higher data rate will not be as robust.

The method being considered in ATSC Subcommittee T3/S9 Candidate Standard CS-110 encodes sufficient data in the 19.39 Mbps ATSC stream to allow a modified ATSC modulator at each transmitter site produce the same datastream using data encoded in the ATSC transport stream and a time and frequency reference such as GPS.


ATSC CS-110 describes a Distributed Transmission Adapter (DXA) used to generate the two synchronization signals multiplexed into the ATSC transport stream before the STL transmitter. A Cadence Signal (CS) establishes the phase of the ATSC data frames relative to the transport stream data packets. The CS is generated by inverting the phase of the MPEG-2 transport stream packet sync word once every 624 packets. Before ATSC modulators can be synchronized, additional information is needed. That information is transmitted using Distributed Transmission Packets (DXPs).

DXPs are not as simple to explain. The DXPs are transmitted using the ATSC Operations and Maintenance Packet Structure. The packets include a header, trellis coder state data, transmitter mode control data, transmitter timing control data and the Reed-Solomon error correction coding. The ATSC candidate standard also allows for other data, such as a transmitter identifier address, network identifier address, and transmitter power control data.

You may wonder how transmitter identification data can be sent if the datastreams have to be identical. A technique known as "buried spread spectrum" is used to put an "RF watermark" on the signal. Ordinary receivers can't detect its presence, but special monitoring equipment can. While this technique does slightly degrade the Additive White Gaussian Noise (AWGN) performance of the receiver, the effect is minimal if a large "bury ratio" is used. The ATSC candidate standard document provides a table showing the impact of bury ratio on AWGN. For example, if a 30 dB bury ratio is used on a transmitter with a base S/N ratio of 30 dB, the transmitter S/N drops to 27 dB and receiver threshold increases from 15.04 dB with no identifier to 15.18 dB. However, if the bury ratio is reduced to 21 dB, the transmitter S/N ratio drops to 20.5 dB and the receiver threshold increases to 16.3 dB.

ATSC recommends the bury ratio be selected so the S/N of the transmitter does not drop below the 27 dB specified in standard A/64A.

I do not have space this month to cover the ATSC candidate standard in detail, but it is available on the ATSC Web site at Click on the reference to "Synchronization Standard for Distributed Transmission".

(click thumbnail)
Spectrum from two combined/slave DX transmitters


Axcera has built a DTV modulator that incorporates the ATSC candidate standard. This modulator design will be used in the WPSX-DT system mentioned last month.

The photo below shows the IF spectrum from two combined DX slave modulators. According to Rich Schwartz, director of marketing for Axcera, "the fact that there are periodic notches in the signal demonstrates that the slave transmitters have synchronized their trellis coders and sync insertion and are transmitting the same symbols. The period and depth of the notches, respectively, are functions of the time displacement and amplitude differences between the slaves. This signal was successfully demodulated by an older Zenith receiver."

Your comments and suggestions are always appreciated. Drop me an e-mail at

Want to read more stories like this?
Get our Free Newsletter Here!