By: O. Bendov

Introduction
In 1948 the FCC issued a "freeze" order, suspending new and pending applications for construction of television-broadcasting facilities pending further study of channel allocation and the method of service and interference prediction. The "freeze" was lifted in 1952 as part of the then FCC's 6th Report & Order, which included the NTSC table of channel allocation and propagation curves. Thereafter, the UHF propagation curves were deemed inaccurate. It took 271 engineers three years, from 1956 to 1959, to complete the experiments leading to the establishment of the propagation curves now in use for NTSC stations. That task was carried out by the Television Allocation Study Organization (TASO), set up by the FCC.

Fifty years later, in 1998, the FCC issued another 6th Report & Order (an ironic coincidence!) that included the channel allocation table for digital television. Unfortunately, the Longley-Rice (LR) propagation model, as used by the FCC for interference and coverage was not validated prior to the issuance of the allocation table. To date, there is no known concerted effort to validate the LR model. The validity of the LR model can be questioned on several grounds. Among the questions are -- will history repeat itself and will a new TASO have to be convened?

Chances are, history will repeat itself Unacceptable interference by HDTV stations to cochannel and adjacent channel NTSC stations have been reported. Mandatory standards for minimal receiver performance do not exist. The threshold level by multiple interferers with a single victim remains unknown. Indoor reception is uncertain. And, the few field tests, limited in scope and with widely varying and unexplained results, have not answered some critical questions. One critical question, fundamental to channel allocation, interference and service prediction, is the concern of this paper.

Examples
A. Channel 53 in Charlotte. NC
Field testing of the Grand Alliance's HDTV transmission subsystem was conducted during 1994 in Charlotte, NC. The terrain was variable. The radiation center was approximately 1940' above mean sea level on a tower 1337' above ground. The Effective Radiated Power (ERP) was set at 31.6 kW.

The expected coverage calculated using the LR propagation model with the parameters recommended by the FCC is shown in Figure 1. The circled locations are the locations were HDTV reception had failed. The radio horizon map is shown in Figure 2. This map provides a more realistic prediction of coverage than the LR model. To match the coverage shown in Figure 1 with that of Figure 2, the percentages of locations and time availability would have to be raised from 50,90 to 99,99 and the confidence margin would have to be raised from 0 to 10 dB.

Click Here for Figure 1

Click Here for Figure 2

1.Test data is published in the interim report (September 1998) of the Model HDTV Station Field Test Program. I am grateful to V. Tawil of AMST Inc. for providing the data on a disk.

2 The field strength of HDTV cannot be measured directly. It is calculated from the total power in 6 MHz and does not include the effective loss of power due to multipath. For a detailed discussion of this point, see "Predicting HDTV Coverage," Broadcast Engineering, March 1966.

Click Here for Figure 5

B. WHD. Channel 30. in Washington D.C.¹
The model station, WHD, operates with an Effective Radiated Power (ERP) of 440 kW (56.43 dBw) from an antenna 405 feet above ground. The measurements analyzed here were performed within the arc over which the ERP is relatively constant. The coverage predicted by the LR model is shown in Figure 3. The circles in Figure 3 are the locations were HDTV service had either failed or is deemed unreliable. Failed or unreliable service locations are defined here as those sites with carrier-noise margin of < 1 dB. The percentage failure was 32% (84 of 263 outdoor sites).

Click Here for Figure3

Figure 4 shows the failed sites plotted on the radio horizon/shadow map. While the radio horizon map may be an improved indicator of coverage over the LR(50,90), it is useless for interference analysis.

Click Here for Figure 4

In all of the failed sites, the measured "Field Strength²" of the HDTV signal was well below that predicted by either LR(50,90) or by the F(50,90) curves as shown in Figure 5. Clearly, a "Field Strength" of 41dBu cannot serve as reliable predictor of coverage.

Even with higher "Field Strengths," service has failed in many locations. As shown in Figure 6, service has essentially failed at 94% of the locations in which the "Field Strength" did not exceed 49 dBu. The difference between the number of sites visited and the number of failed sites begins to spread for "Field Strengths" > 49 dBu.

Click Here For Figure 6

Evaluating the coverage of just two stations is hardly definitive, but the results suggest that the LR(50,90) propagation model may not be generally applicable for reliable HDTV coverage prediction unless its input parameters are properly adjusted. Similarly, the usage of LR(50,10) for interference analysis may need reevaluation. The shortcomings of the LR model are:
á Unrealistic statistical margins.
á Multipath, the most serious cause of HDTV service failure, is ignored.
á The LR algorithm applies to a single carrier, not to wide band (6 MHz) signals.

Realistic Propagation Modeling

In producing the HDTV channel allocation table, the FCC relied, simultaneously, on two different propagation models. One model was the empirically derived curves for NTSC. The second was the LR. From a technical standpoint, there would be no reason for mixed use of different models, with widely divergent prediction of service, if one of the two were appropriate for the purpose.

A mathematical propagation model requires that the user enter several critical parameters before calculation can proceed. As is always the case with computer modeling, the axiom "garbage in - garbage out" applies. What are the critical parameters and what values did the FCC assign these parameters?

Height of Receive Antenna
The median gain/loss of signal due to a change in the height of a UHF receive antenna from a reference height of 30 feet is³:

H is the height in feet, A=4, 6, and 8 are respectively, for Rural, Suburban and Urban areas.

³ ITU Recommendation P.370-7, 1995.

The FCC set the antenna height above ground at 30'. That elevation may have been appropriate during the 1950's. Nowadays, the average height of outdoor antennas would be lower. For example, if the height of the receive antenna were 15' above ground, a received UHF signal would be -6.0 dB below that expected from an antenna 30' above ground. It would take quadrupling of the transmitter power to make up this loss.

Receiver Noise Figure

For UHF-NTSC receivers, the FCC mandates a minimum noise figure of 14 dB. A minimum noise figure specification for HDTV receivers has not been mandated. The FCC has used a noise figure of 7 dBs a planning factor for UHF-HDTV but that noise figure is not binding on the manufacturers of HDTV sets.

Even if the manufacturers specify 7 dB as a noise figure, that noise figure cannot be used for service prediction without further modification. The reason for that is that the factory noise figure is measured with the receiver ("load") matched to the noise generator ("source"). A household antenna is rarely matched to the receiver. A Standing Wave Ratio (SWR) of 5:1 across the UHF band is not unusual and a ratio of 2:1 is common. The effective noise figure increases for a mismatched antenna/ receiver. For a 2:1 mismatch, the effective noise figure increases by 3 dB over the factory's noise figure. It would take more than doubling of the transmitter power to make up this loss.

The factory specification for noise figure of production run UHF-HDTV receivers will probably be closer to 10 dB. That is 3 dB higher than that specified in the planning factors. The effective noise figure, accounting for the SWR in the downlead cable, would then be closer to 13 dB than to 7 dB. All these shortfalls could be mitigated by a "smart" receive antenna⁴

⁴First proposed by the author at the PS/'WP3 meetings and later elaborated in two papers presented by the author at the 1994 and 1997 NAB conventions.

The Effective Earth Radius Multiplier Factor

The variations in propagation conditions require that the US be divided into three zones. The 1959 TASO report recommended that the multiplier factors for the three zones be:

Therefore, the radio horizon, which is the line-of-sight for UHF waves, is not uniform around the US. Based on effective earth radius recommended by TASO, the radio horizon for various antenna heights around the US are:

The FCC's LR model assumes an effective earth radius of 5280 miles everywhere.

Statistical Margins
The received power level of over-the-air transmission depends on factors that are neither constant nor exactly accountable for. Local terrain and weather variations are examples. Therefore, statistics and probability are used to supplement the calculation of the received signal level. With a statistical description of measured data at hand, a safety margin is assigned to the unpredictable degrading factors. The margin, expressed in dB, is a power-loss penalty subtracted from the ERP to ensure that the statistics of HDTV reception would meet at least the preset levels.

The LR model allows for an assessment of most but not all of the required margins. Included in LR are margins for location and time variations and confidence level. Not included in LR, but a necessity in digital RF links, is a margin for multipath.

a. Reliability margin
This margin, in dB, is subtracted from the median signal level to insure a higher probability than having the desired signal at just the best 50% of the location at least 50% of the time. What the correct percentages are for digital television where the picture abruptly disappears rather than fades, no one knows. The FCC has set the same reliability statistics for HDTV as those in use for NTSC. That is, HDTV is considered to be reliably available if it can be viewed at least 90% of the time at the best 50% of the locations. At the other 50% of locations, any percentage of time availability would be acceptable. A loss of picture 10% of the time within the best 50% of the locations would be acceptable as well.

b. Confidence margin
The LR model requires as input of the confidence level that the designed reliability (time and location) will be met by a percentage of broadcasters in any one market. The FCC set the confidence level at 50%. That means that at least 50% of the stations in any one market will have HDTV signals available at least 90% of the time at the best 50% of the locations. The signal availability statistics of other 50% of the stations in the same market may be lower.

It has been pointed out⁵ that the result of using the LR model with a 50% confidence level is not comparable to that of using the FCC's (50,50) curves, and that confidence levels of at least 90% must be used in the UHF band.

⁵ Louis A. Williams, Jr., private communication sent to the Association of Federal Communications Consulting Engineers (AFCCE) DTV Committee dated August 2, 1997.

c. Receiver equalizer imultipath margin
Experience to date has shown that multipath propagation is a major detriment to HDTV reception, even when received by outdoor antennas. The effect of multipath propagation appears as in-band amplitude and phase distortion. If the multipath does not exceed a certain time delay and/or certain magnitude, the equalizer at the receiver will provide the correction - at a price. The price, in dB, amounts to a power penalty that effectively increases the threshold level of the carrier - noise ratio.

The use of a multipath ("dispersion") margin in wide band digital radio links is well known and thoroughly documented⁶. The addition of this safety margin is important because the LR model is presently designed for single-frequency signals only.

⁶ For example, "A Simplified Method for Prediction of Multipath Fading Outage of Digital Radio" by Y. Serizawa and S. Takeshita, IEEE transactions on Communications, Vol. COM-3 1, August 1983. Also, EDX Engineering MSITETM manual.

Interference Prediction
The assumptions in the FCC's Planning Factors are that the outdoor receive antenna is 30 feet above ground and that the antenna's backlobe (-14 dB @ UHF) is oriented toward the interfering station. But what about cable head-ends with receive antennas 300 feet above ground? What if the home user cannot rotate the outdoor antenna in such a way that the backlobe is toward the interfering station? What if rotating the antenna to minimize the interference also causes a loss of the desired channel? These questions may have to be faced in the months to come as the levels of cochannel and adjacent channel interference into NTSC channels becomes clearer⁷. Here too, a validation of LR(50,10) may be timely.

⁷As of November 1, 1998, unexpected cochannel interference has been reported in WI and adjacent channel interference has been reported in PA.

Excessive interference could also result from higher than planned ERP in some directions. There are two potential causes for excessive ERP:

á Filing based on an incorrect antenna gain.
Until November 1998, licensing applications for HDTV used form FCC 302 which did not require that the antenna's RMS and peak gains both be listed. The peak gain can exceed the RMS gain by several dB. Since the Commission's rules are not specific on this matter and RMS gain has been used for omnidirectional NTSC stations, it could be assumed that either gain is acceptable. The relation between the two definitions of gain is shown in Figure 7.

Click Here for Figure 7

The peak gain of the omnidirectional antenna pattern is 3.25 dB above the RMS circle. Therefore, if filed on the basis of RMS gain, the antenna whose pattern is shown in Figure 7 could be used as part of a checklist application and licensed for ERP of 473 kW (RMS) which would not exceed the allowable ERP anywhere. However, the peak ERP of that antenna would reach 1,000 kW, exceeding the allowable ERP in two directions.

In November 1998, a new form, FCC 302-DTV, was issued. The new form requires that both peak and the RMS gains of the antenna be specified. That requirement closes only half of the loophole. For example, antennas may require a larger diameter support pole in order to meet certain structural specifications. Now, should the RMS and peak gains of the antenna assuming the supporting pole does not exist be submitted or should the as-installed patterns be submitted?

á As-installed antenna pattern.
Antennas that are side-mounted on the tower shaft and antennas that are placed next to one another on one tower will have their patterns modified by the proximity to other conducting objects. An example of the modified pattern of an omnidirectional antenna, side-mounted next to a tower with an 8-feet face, is shown in Figure 8. Clearly, in some directions the ERP has increased by as much as 3.64 dB.

Click Here for Figure 8

The potential problems that could arise from the situations described are easily correctable. First, license applications should include the antenna patterns and gain specifications with the support poles (if used). Second, for antennas installed within 50 feet of a metallic obstruction, factory patterns as well the as-installed patterns would be required as part of the license application.

Conclusion
Analysis of the available field test results coupled with key theoretical considerations shows that a modification of the LR model will be required before it could be effectively used for HDTV coverage and interference prediction⁸.

There may also be software code implementation error. Hammett & Edison, a consulting engineering firm, has asserted to the FCC that, on average, 18% of the population reside in cells arbitrarily assumed to get HDTV service even though a proper LR(50,90) calculation could not be performed in those cells.

This paper has also demonstrated that a "Field Strength" 41 dBu is inadequate and inappropriate as a measure of HDTV service contour and as a measure of service within that contour.

The limits to reliable HDTV service will not be known until several thousand sets are in place in each of the major markets. In the meantime, the radio line-of-sight could serve more reliably as a predictor of HDTV coverage than LR(50,90).

Several steps should be taken now to help smooth the transition period ahead:

First, the LR model should be modified for wide-band signals, its statistical margins adjusted and multipath margins for urban and suburban areas added. The multipath margins would be related to the performance of the receiver's equalizer.

Second, future field tests should be expanded for the purpose of determining the correct parameters used by the LR model for coverage and interference predictions.

Third, the FCC should establish minimum performance standard for HDTV sets. The standard should cover noise-figure, selectivity and equalization.

Fourth, the FCC, in cooperation with industry and academic institutions, should establish TASO II as an advisory group to help resolve key transmission and allocation problems related to digital television broadcasting.