FCC Spectrum Analysis Doesn't Add Up
This week the FCC released a paper, Spectrum Analysis: Options for Broadcasting OBI Technical Paper No. 3
. I have great respect for the engineers whose names are on the report. During the 40 years I've been involved with broadcasting, FCC reports were, with few exceptions, unbiased, accurate and complete. That is why upon critical reading I was so disappointed to find this attractively presented Paper to be biased, incomplete, and, in some ways, inaccurate. Critical information is missing and conclusions are presented as fact when data doesn't support them. The analysis glosses over data that refutes the National Broadband Plan's (NBP's) recommendation for taking away broadcast spectrum and makes unjustified assumptions to support them.
I'll highlight a few examples, but urge you to read the document and draw your own conclusions.
The Paper notes that the FCC staff and contractors have been developing a new Allotment Optimization Model (AOM) to achieve the NBP's goal of taking 120 MHz from the 294 MHz (240 Mhz) after removing three channels for land-mobile and television Channels 2 through 6 currently assigned for TV broadcasting. Not surprisingly, when the model was used to see how much spectrum could be recovered by repacking existing full power stations, only 6-42 MHz of spectrum was found to be available for reallocation, depending on inclusion or exclusion of border restrictions.
The scenario did not move UHF stations to VHF or VHF stations to UHF. Cellular technology is briefly discussed but the only supporting data is Appendix A, which includes a graph showing 650 MHz vs. 1900 MHz rural cell areas plotted against maximum allowable path loss in dB. It's an example of the document's tendency to present impressive looking data with little relevance to the real world. In this case, multipath and self-interference are ignored, and the table is intended for rural areas, where there is less demand for both wireless spectrum and broadcast spectrum.
Appendix D is a cut and paste of parts of FCC OET Bulletin 69, with half a page devoted to a now out of date table showing desired-to-undesired interference ratios for analog TV. A close look indicates that only three lines apply to DTV-into-DTV interference.
Before you argue that many analog LPTV and Class A TV stations are broadcasting analog, that doesn't matter in this report.
There is a small section on LPTV stations, but as far as I can tell, they were not included in the studies. The Paper repeats the NBP suggestions for ways to accommodate LPTV stations, but notes: "To the extent that a reallocation compresses the broadcast TV bands, non-Class A LPTVs may be forced to move, and therefore incur relocation costs, and they may find fewer available channels slots which they can occupy."
Regarding wireless microphones and TV white space devices, the document says they must be considered, but states, a consideration of any potential effects and solutions would be "beyond the scope of this paper," as it is intended to address "particular considerations" that apply to full-power TV. In my opinion, with no available TV channels where most people live, the options for LPTV stations and wireless microphone users would appear to be limited at best.
In fairness to the authors, there is some relevant and useful information in the document's other Appendices. As an example, "Appendix C: Optimization Model Formulation" provides details on the AOM used to determine the amount of spectrum reclaimed from repacking and channel sharing.
One thing I found interesting was that "the model does not incorporate terrains conditions in order to determine the minimum allowable spacing between stations." Users can selectively relax spacing restrictions between pairs of stations in a DMA "based on an assessment of terrain or other considerations." No details are provided on how that would be applied objectively.
Based on the results from the AOM repacking study, the Paper suggests one way to clear half the broadcast channels would be to put two stations on one channel. Exhibit G shows the result of this study. If border restrictions are ignored, the Exhibit indicates that 120 MHz of spectrum can be taken from broadcasting if 204 stations share a channel. That may not sound like much, but is an example of where insufficient data is presented for readers to get a true understanding of the impact.
Only 12 percent of the country's stations sharing channels doesn't sound like much, but the percentage would have to be higher in markets like Los Angeles and in areas in the northeast where some stations had trouble finding a suitable channel for DTV after channels 52-69 were reallocated. In the scenario, some 707 stations (41 percent) would have to change channels. In addition, more than half of the stations would lose some service area.
What is interesting is that if the amount of spectrum taken from broadcasting is reduced to 72 MHz⎯enough to protect active border allotments⎯only 38 stations would have to share a channel and another 392 would have to change channels.
Without more data it's impossible to determine where those 38 stations would be located. The exhibit shows that the portion of the viewers that would gain service from stations improving coverage is larger than the number of viewers losing service from stations with reduced coverage.
Again, the limited data makes it hard to visualize what the real impact is. The stations gaining population may be those with limited power now, while the ones likely to be losing population could be the higher power, major market network affiliates⎯the ones also likely to have the most viewers.
Look at these columns again, the "Without Border Restrictions" numbers are a red herring
. It seems unlikely the FCC would be able to ignore active border allotments when repacking the TV band. Unless the Commission is suggesting that 120 MHz may be too much to take from TV broadcasting⎯which doesn't seem to be the case⎯the "Active Allotments" column is also a red herring.
Where will the other 48 MHz come from?
Conveniently missing is information on the shortage of channels, the greatly increased loss of service or the additional channel sharing that would be required to reclaim that extra 48 MHz. Why isn't that analysis included in Exhibit G?
Perhaps it's not there as it would conflict with the statements elsewhere from the FCC that broadcasters will not have to give up spectrum involuntarily, and that taking away 120 MHz of TV spectrum will not prevent them from multicasting or transmitting Mobile DTV.
Indeed, the amount of space—with examples, of how even network affiliates broadcasting HDTV can share one channel—ignores Mobile DTV or multicasting considerations in channel sharing. The scenarios described for combining two stations on one channel with HD can work only if some—perhaps all multicast channels are dropped and Mobile DTV is not transmitted.
The data rate required to transmit video programming, as the document shows, is highly dependent on content. Statistical multiplexing can be used to dynamically share bandwidth, allowing maximum spectrum efficiency.
In the case of ABC (one of the examples cited), I think that anyone who's watched ABC's second HD channel would agree its quality is significantly impacted by what's airing on the main channel. How would that work in a shared channel arrangement?
If you look at the situation with regard to WRC-TV in Exhibit L, you can see how statistical multiplexing works. Programs are given different priorities. When NFL Football needs more bandwidth, it can take it from the two multicast channels. When Universal Sports (the "third channel") needs more bandwidth it can take it from NBCPlus.
Viewers on Universal Sports will see some artifacts (and there have been complaints) when the HD channel needs the bandwidth to avoid artifacts. The second channel is primarily used for weather maps and talking heads, and as you can see, doesn't require much bandwidth.
Now imagine trying to squeeze two HD channels into 19.39 Mbps with mobile and multicast. Mobile DTV requires a fixed bandwidth as does audio and PSIP. If video description is required in the future, that will take additional fixed bandwidth. The lowest bitrate for HD shown in the paper is 8 Mbps. Double that, add another 1.2 Mbps for PSIP and Dolby 5.1 audio for the two programs, and you've used up 17.2 Mbps. The 2.19 Mbps left over is enough for one Mobile DTV program or perhaps two low-action multicast streams (don't forget they need fixed bandwidth for audio). Which of the stations sharing the channel gets the single Mobile DTV channel?
I've seen tests using the latest video encoding techniques available⎯which I would argue match or exceed the improvement shown in Exhibit I from Tandberg. These, by the way, according to the footnote are based on a study of "SD streams extrapolated for HD," and not HD.
The tests I've seen indicated that with challenging sports material running at 1080i, the video data rate had to be greater than 9 Mbps to avoid objectionable artifacts. While it might be possible to squeeze in low action HD content together with sports in 19.39 Mbps, if two stations are sharing a channel which one has priority?
Exhibit N, "Hypothetical Coincident Programming Snapshot Based on Pairing of Eight Stations in Washington, D.C." is also very interesting. Look at the third item—NBC-Uni. This is based on the premise that Univision is transmitting SD. Univision is now broadcasting in HD. Imagine how unhappy all of those World Cup viewers would be if Univision was limited to SD-only, due to channel sharing. NBC doesn't do well either.
Even if Univision operates in SD, NBC would have to drop Mobile, its popular "Universal Sports" multicast, or—more likely—both, since NBC couldn't take bits from Univision when needed for HD.
The Paper's analysis doesn't allow stations to move from SD to HD, as Univision has done, nor does it accurately reflect that on a shared channel it's likely to be difficult, if not impossible, for one station to steal bits from the other when needed. The analysis should have been based on peak bit rate, not average bit rate. Some sharing is possible, but I think it's disingenuous to suggest that it can be done without affecting channel choices and Mobile DTV.
I haven't addressed channel allocation issues in this article, but if you're concerned about the future of broadcasting, look at Section IV, Reallocation Mechanism. It says that with a repacking scenario, "every existing broadcast TV license would go into a 'pool' for determination of channel assignment post-auction."
As noted at the beginning of this article, I was very disappointed in this report. I agree there are ways to allocate at least some broadcast spectrum for wireless broadband, although the amount varies by location. Also, moving to MPEG-4/AVC compression could provide significant reductions in the amount of bandwidth required for HD content although at the risk of disenfranchising existing viewers. Would we need a DTV converter box program V2.0?
If the OBI Paper was less of a cheerleader for the NBP, and more of a detailed technical analysis of spectrum availability by market and a discussion of alternatives, it could have facilitated discussions in the upcoming Broadcast Engineering Forum.
It doesn't help to list potential problems and then ignore them or to dismiss two of the most promising benefits of the DTV transition⎯more choice in programming and Mobile DTV⎯in favor of an economic analysis that shows using spectrum for TV broadcasting doesn't pay. Multicast programming has allowed unique programming to be provided to special interest groups, such as cycling on Universal Sports or ethnic programming in a wide variety of languages, as can be seen on multicast channels in Los Angeles. The low cost of multicasting makes this possible.
Is it really right to argue that just because that Korean multicast programmer may not pay as much for access to her audience as Verizon would for the same spectrum, this programming has no place on free TV?
PBS doesn't make a lot of money, but I'd sure hate to see it, or its new multicast channels, disappear.