I recently participated in a panel discussion at a symposium held in Chicago by the IEEE Communications Society on the topic of "Opportunistic TV White Space Re-Use: Facts and Fiction." In preparation for this I analyzed the interference issues, which I will present to my readers here. Perhaps the FCC will propose a set of technical rules for such sharing even before you read this.
I studied the table of DTV channel allotments given in FCC 07-138 Appendix B at the nation's second largest DMA—Los Angeles—and looked for "vacant channels" in each TV broadcast band. All low VHF channels can be considered "vacant" in the sense that there are zero low VHF band DTV allotments in the Los Angeles area. In the high VHF band there are allotments of channels 7, 9, 11 and 13, so one might think that channels 8, 10 and 12 are vacant. Well they are not allocated, but each is adjacent to one or two local allotments. Sideband splatter from DTV transmitters on these channels may interfere with reception of data signals on channels 8, 10 and 12.
In the UHF band there are unallocated channels, however channels 14, 16 and 20 are already being shared in Los Angeles with public safety. There are scads of unallocated channels left, but every one of them happens to be adjacent to one or even two local DTV allotments. So in this market, all vacant channels in the high VHF and UHF bands are also adjacent channels to one or two broadcast channels. So are adjacent channels really suitable for sharing?
DTV INTERFERENCE FROM DTV SIGNALS
Let's assume there is a DTV allocation next door for 1,000 kW ERP in the horizontal plane. The maximum field strength of this DTV signal at 1 mile from the tower is 132.8 dB above 1 microvolt per meter (dB µV/m). On the ground, its field strength will be somewhat lower, depending on the vertical directivity and beam tilt of the broadcaster's antenna. At 10 miles this field strength will be down by 20 Log d (miles) or 112.8 dB µV/m.
In each channel adjacent to a DTV channel, there will be radiated DTV sideband splatter 44.2 dB below the DTV signal power being radiated within the DTV channel. The field strength 10 miles from the broadcast tower of radiated sideband splatter in the channel the unlicensed facility may choose to use will be 68.6 dB µV/m (112.8 dB µV/m – 44.2 dB). This is the noise floor in each channel adjacent to the broadcast channel.
The FCC speaks of peak power for unlicensed devices and the average power as for DTV signals. Our ATSC signal is rather benign compared to COFDM modulation with respect to its peak-to-average power. I will assume that unlicensed signals have our modest 6 dB peak-to-average power ratio, so 4 watts peak power for "fixed facilities" is 1 watt average power. The field strength of a 1 watt signal at 1 mile is 72.8 dB µV/m (102.8 dB µV/m – 30 dB).
What this means is that the data signal-to-DTV sideband splatter power ratio in each adjacent channel is 4.2 dB (72.8 – 68.6 dB). This noise is in the data signal's channel so it is co-channel noise and the SNR = 4.2 dB one mile from a 4 watt (peak) unlicensed transmitter.
Only the most rugged of digital modulation schemes can operate with a SNR of 4.2 dB and those provide very low data rates.
A strong DTV signal may generate third-order intermodulation products which fall in both adjacent channels. This locally generated sideband splatter is co-channel noise for the data signal to have to contend with as this is a broadband noise extending across each adjacent channel.
No one has ever tested data transmission where a strong DTV signal overloaded the data receiver front-end. However, we have lots of data concerning front-end overload of DTV receivers. I expect these data receivers would use the same off-the-shelf DTV tuners designed for the DTV receiver market. Their interference rejection has been documented in a report by the FCC Laboratory: FCC/OET 07-TR-1003 dated March 30, 2007.
I believe that data transmission on adjacent channels to local DTV channels is not technically feasible.
INTERFERENCE FROM A DATA SIGNAL
Let's look at interference from the data (or "D" channel) signal on an adjacent channel into a DTV receiver. The FCC in the above report has noted that 29 percent of the area inside the noise-limited contour of DTV stations has a signal margin of less than 3 dB. This means that the predicted field strength of DTV signals is between 44 dB µV/m and 41 dB µV/m in 29 percent of the station's predicted coverage area. Ouch! There might be people living out there.
At the edge of the noise-limited coverage, the FCC reckons DTV field strength is 41 dB µV/m. The FCC assumed the DTV receiver would generate –99 dBm of noise. Now if another noise source also at –99 dBm is present, the total noise power increases by 3 dB. Let's assume this additional noise in the D channel is from an unlicensed device (U), so U = –99 dBm in the D channel 3 dB inside that contour. That additional noise will block DTV reception in the outer 29 percent of the station's noise-limited coverage area. Ouch, again!
This noise got into the DTV channel by generating third-order distortion products in the DTV receiver's tuner. We have experimental data on this from the same FCC report: Table A-8 gives the threshold U power in dBm for D = Dmin + 3 dB, the new limit on DTV coverage. For an undesired signal on channel N–1, the threshold of interference is –46.3 dBm; for N+1 it is –46.9 dBm. Therefore the data signal into a DTV receiver must be less than this value and this determines how close to the DTV receiving antenna the data transmit antenna can be without causing harmful interference. Thank goodness the FCC measured the threshold U power for desired DTV signal power at Dmin + 3 dB.
My conclusion is that first adjacent channels are unsuitable for use by unlicensed devices from both the broadcaster's and the data forwarder's perspective.
Another conclusion can be reached concerning second adjacent channels N+/–2. Again, the FCC data shows that the threshold of interference at Dmin + 3 dB is –53.1 dB for channel N–2 and –52.7 dBm for N+2. The FCC was surprised at this unanticipated finding that the N+/–2 interference is actually worse than N+/–1 situation by about 6 dB and that ain't hay is it?
There is a sound technical reason for this unanticipated result. DTV receivers available in the 2005/2006 timeframe appear to have wideband RF AGC (automatic gain control). Analog TV sets always sensed the signal power after the IF filter so they responded to the desired signal, but did not respond to undesired signals. Wideband RF AGC receivers can reduce the gain of the RF amplifier if they sense a signal on channels N–1, N, and N+1, and to a lesser degree, a strong signal on N+/–2. This is why interference on N+/–2 is more potent than from N+/–1. Perhaps some manufacturers will widen their RF AGC pre-detection bandwidth some day. There are now tens of millions of DTV receivers out there, in dealer's stores, or en route from manufacturers.
One can conclude that N+/–1 and N+/–2 are all taboo, as my friend, Dr. Oded Bendov has been writing for years. Now you know a little more about why.
WHAT ABOUT MULTIPLE BASE STATIONS?
The range of the personal/portable data transmitters folks will be carrying around will be limited by their battery capacity or the FCC limit for their peak EIRP, which may be under 100 mW. Base stations will be numerous throughout each community. Let's assume they are on a grid with 2 miles between base stations. Each data receiver will be within 1 mile of one, two or even four base stations. Base stations will be operating multiple transmitters each on a vacant channel. If two base station transmitters are operating simultaneously on channels 33 and 36, their signals will be received at 1 mile with up to –12 dBm power each. This is more than enough to overload receiver front-ends, so we can expect third-order distortion products to be generated in receiver front ends.
Here I am assuming that data receivers will have front-ends like those of DTV receivers as both would require a frequency-agile tuner which covers the broadcast TV channels. Readers of this column know that with two signals on channels 33 and 36, the third-order IM products fall principally in channels 30 and 39. This noise in channels 30 and 39 generated in the data receiver may prevent it from working on these channels.
A DTV receiver may experience jamming on channels 30 and 39 from data signals on vacant channels 33 and 36. No tests for this sort of interference have been conducted to my knowledge.
Readers also know about triple-beat third-order intermodulation products where three or more undesired signals overload a receiver generating a broad spectrum of noise possibly ruining reception on a much larger number of channels.
CONSIDER POLARIZATION DIVERSITY
Broadcasters are permitted to radiate as much power in vertically polarized waves as their authorized power, which is specified in a horizontal plane. Will they do it? They will have to radiate a substantial amount of power with vertically polarized waves for personal/portable and vehicular reception. Vehicular antennas are usually a quarter wavelength vertical rods above the rooftop which serves as a ground plane for the vertically polarized waves. Mobile DTV seems to be all the rage these days so broadcasters won't give up this right.
But on the other hand, users of personal/portable data transmitters cannot be held accountable if they do not keep their antenna in a vertical plane.
Furthermore, TV signals—especially UHF signals—tend to suffer depolarization.
While most DTV signals are being radiated with 100 percent power in a horizontal plane, depolarization does take place and we observe in the field that there is no plane of the receiving antenna in which we find a null in received signal.
IF THERE IS A SOLUTION, WHAT IS IT?
After Feb. 17, 2009, data folks could use some of the virtually vacant low VHF channels. Yes, there will be 38 DTV stations on channels 2–6. I believe that nearly all of these could find a suitable channel outside of the low VHF band if they look hard enough.
Fixed location unlicensed transmitters would probably be allowed 4 watts (peak) EIRP on any vacant TV channel they wish to use. The range of a 4 watt signal depends on its frequency. Remember the term "dipole factor?" The difference in dipole factors from 69 MHz to 615 MHz is 19 dB. That is almost 100:1, so a 4 watt signal on a low VHF channel should be equal to a 400 watt UHF band signal for range. UHF signals do not tend to follow the earth's curvature, while signals below 100 MHz do. A 4 watt signal can reach beyond the optical horizon reliably. So, if the customer is living in a sparsely settled area where long distance telephone charges are prohibitive for Internet access, he really can benefit from using these low VHF channels. Such rural users will gravitate to the low VHF band automatically, I believe. In fact, many of these folks already have a big antenna atop a tower, so what is the problem with their using it for wireless Internet access?
As for TV reception, they will now need a high V plus UHF antenna for that application. They will also need a high pass filter to keep the low VHF data signals out of their DTV receiver.
Stay tuned, more will follow next month on the white spaces issue.
Charlie Rhodes is a consultant in the field of television broadcast technologies and planning. He can be reached via e-mail at firstname.lastname@example.org.