This month I'll look at a few more papers presented at the 2011 IEEE Broadcast Technology Symposium, two on coverage prediction and one on a new technique for adding data to an ATSC 8-VSB signal without reducing the 19.39 Mbps data rate.
PREDICTING COVERAGE
Broadcasters developing business plans for Mobile DTV using ATSC A/153 need a reliable coverage prediction model. This is a challenge because receivers are likely to be used indoors, on streets with building shadows and in cars through varying environments.
The Open Mobile Video Coalition, through its technical advisory group (OTAG) has been conducting field measurements of Mobile DTV signals. NPR Labs research on mobile/portable reception for FM IBOC can also be applied to ATSC Mobile DTV.
At the symposium, John Kean from NPR Labs presented "Parameters for Coverage Maps of ATSC Mobile/Handheld Television." The accuracy of terrain elevation models and land use land cover (LULC) data has a direct impact on the accuracy of coverage models. Kean explained that the LULC classifications used by the FCC are not designed for propagation studies as many categories are related to land use or agriculture; others are ambiguous ("developed open space"); and some are completely unsuitable—highways are designated "High Intensity Urban." NPR Labs created its own land cover data (NLCD) to correct these problems.
NPR Labs is now working on applying appropriate attenuation values based on its NLCD to the Telecom Industry Association's (TIA) Bulletin TSB-88 data. Early analysis with measurement data found that Terrain Integrated Rough Earth Model (TIREM) doesn't behave well with high-resolution terrain data. The best match was TIREM and "Globe" 1 km elevation data.
NPR Labs uses TIREM because they found ITM ("Longley-Rice") greatly over-predicts signal levels and misses nulls. Work to refine their coverage model involves field measurements to determine the optimum adjustments to the model for frequency, polarization, receiver height, etc.
Bill Meintel presented the results of his research into ATSC Mobile DTV coverage prediction in the paper "Service Prediction Modeling for ATSC M/H." Like Kean, he found TIREM provided results that more closely matched field measurements than ITM, but it wasn't the final answer.
In addition to path loss, accurate coverage prediction also needs to include the receive site location (inside or outside?), interference (man-made and naturally occurring) and target receiver capabilities (service threshold and interference rejection).
In a mobile/handheld environment, Meintel found TIREM over-predicted signal strength by 19 dB and Longley-Rice over-predicted by 25 dB for more than 95 percent of the points studied. However, add FCC land-use clutter factor to Longley-Rice and the difference with TIREM goes away. In reality, neither model was very good at matching actual field values.
Meintel didn't have a magic solution, but had some ideas to maximize the match. One is to use higher-resolution terrain data—1 second or better. Unfortunately, existing terrain databases are not very accurate. SRTM has been suggested as an option because it picks up buildings and forest canopy, but when Meintel compared SRTM data to known structures he found that, while its elevations were above average terrain, they were not close to actual building heights.
Furthermore, while STRM data or actual building height data can be used in coverage studies, Meintel explained buildings do not have the same impact on the signal as terrain and the impact varies depending on building type. His completed work showed building penetration loss ranges from 5-20 dB with an average of 12 dB and height loss going from 30 feet to 6 feet varies from 5-14 dB with an average of 9 dB.
Much more work is needed to obtain accurate ATSC M/H coverage studies. I'm sure we'll be hearing more from Meintel and associates at Meintel, Sgrignoli and Wallace.
AUGMENTED DATA TRANSMISSION
Researchers from ETRI (Sung Ik Park, Hyoungsoo Lim, and Heung Mook Kim) and Yiyan Wu from CRC Canada presented their paper on "Augmented Data Transmission (ADT) for ATSC" at the Symposium.
Table 3 from FCC Office of Engineering and Technology Bulletin 73, "The ILL R Computer Program for Predicting Digital Television Field Strengths at Individual Locations." It reminded me of techniques for adding digital data to the analog TV signal proposed many years ago before the ATSC DTV standard was finalized. These never saw widespread use. Will this technology be different?
The system builds on A/110B ATSC transmitter ID (TxID) technology in that it uses hierarchical 2-VSB modulation buried in the 8-VSB signal. TxID is extremely robust but provides data rates of 160 bps at most—enough to identify a transmitter in an SFN, but not enough for other uses. ADT adds advanced error correction technology and technology to cancel out the 8-VSB DTV signal to increase the data rate up to 2 Mbps.
The ADT system does not take any bandwidth from the 19.39 Mbps bandwidth used for A/53 and A/153, but there is a cost. Trade-offs for ADT are greater than those for TxID. A 24 dB bury ratio (ATSC SNR of 24 dB) allows a 1 Mbps ADT signal with 10E-6 BER at an 18 dB receive SNR. Increase the bury ratio (BR) to 30 dB (negligible impact on ATSC reception), and a 23.5 dB SNR is needed for reliable ADT demodulation at 1 Mbps. More than 26.5 dB SNR was required at this BR for a 2 Mbps ADT rate. The paper provided no data on how well ADT would work in a mobile/handheld environment. The high receive SNR requirement broadcasters are likely to accept would appear to limit it to fixed applications. Perhaps additional coding would allow a robust 500 kbps data rate for mobile/handheld service at a BR of 27 or 30 dB.
I welcome your comments and suggestions! Email me at dlung@transmitter.com.
