Progress in the Debate over TV White Space
Tuesday (July 11, 2017), Microsoft unveiled their current vision for unlicensed radio services in the TV White Space (see “Microsoft calls for U.S. strategy to eliminate rural broadband gap within 5 years”.)
Microsoft’s proposal demonstrates progress in the debate over use of the TV White Space. A decade ago, proponents of unlicensed operation in the TV White Space stated that it “will transform every aspect of civil society.” (No cite given because I don’t want to be cruel. But, search engines are your friend.) Yesterday, Microsoft made available a white paper about White Space that asserted that “Overall, TV white spaces technologies appear to be the optimal solution for a little more than 19 million people . . .”
Nineteen million people is a tad less than 6% of the population. Toning down the rhetoric in support of unlicensed use of the TV White Space from “transform civil society” to “provide a more cost-effective alternative for 5.8% of the market” is real progress. (The quote regarding 5.8% is my statement, not Microsoft’s.)
Still, I doubt that unlicensed operation in the TV White Space would “provide a more cost-effective alternative for 5.8% of the market.” The Microsoft white paper asserts that using TV White Spaces could provide access to 23.4 million people in rural areas for a cost in the range of $10–15 billion; it also asserts that using commercial wireless in the 700 MHz band would cost $15–25 billion to provide the same coverage. (Microsoft White Paper at pp. 12-13.) Thus, Microsoft asserts that using 700 MHz spectrum is projected to cost about 1.5 times as much as using the White Space. Microsoft cites a study by the Boston Consulting Group for these cost estimates; I could not find a copy of that study or a description of its methodology.
However, the Microsoft white paper drops some hints about the methodology. Hint number 1: they note that most TV White Space operations will be on TV channel 37 (608–614 MHz) or in the duplex gap (652–663 MHz).
Hint number 2: they state (incorrectly—more about that later) that a TV White Space signal can travel four times the distance of a 2.4 GHz Wi-Fi signal. Well four times the distance is exactly what one would calculate if one used the Friis Transmission Formula to compare signal strengths and assumed that the TV White Space system operated at 600 MHz and the antennas in both systems had equal gain. Microsoft’s assertion that signals in the TV White Space travel four times as far as 2.4 GHz Wi-Fi signals would not apply if the Wi-Fi system had an outdoor antenna as big as the antenna of the TV White Space device; here’s an example of such an antenna.
Using a big antenna like that the Wi-Fi signal received at the residence would be just as strong as the comparable White Space signal; the uplink Wi-Fi signal from the residence to the base station would be much stronger than the White Space signal.
But the Friis formula may explain the difference found in the Boston Consulting Group study between the cost of rural coverage using TV White Space and the cost of using 700 MHz LTE. If one were to apply the Friis transmission formula to compare the coverage of a 700 MHz LTE system with that of a 600 MHz TV White Space system in the same way that Microsoft applied it to Wi-Fi versus a TV White Space system, one would calculate that a base station in a TV White Space system would have about 1.4 times the coverage area of a base station in a 700 MHz LTE system.
If rural America could be covered by 10,000 TV White Space cells, it would require 14,000 700 MHz LTE cells to give the same coverage. This simple calculation leads to a conclusion (national coverage using 700 MHz systems would cost 1.4 times more than would using TV White Space systems) that is essentially the same as the Boston Consulting Group study’s conclusion that using 700 MHz systems would cost about 1.5 times more.
The analysis presented above is incorrect for at least two reasons. First, the Friis Formula is probably not the appropriate propagation model in this context. Second, even if it were the correct model, the conclusion would be wrong because of an omitted factor. Wireless coverage depends on both signal strength and bandwidth. Wireless carriers have access to 84 MHz of spectrum in the 700 MHz band—more than four times the 18 MHz of white space that Microsoft hopes will be available at all locations. In many circumstances, that extra bandwidth would more than compensate for any slight difference in signal attenuation. Consequently, systems operating in the 700 MHz band should be more effective alternatives than is suggested by Microsoft’s analysis.
A Missing Competitor
One important factor that is not addressed in the Microsoft white paper is use of the 70 MHz of spectrum recently auctioned off by the FCC. This spectrum brackets the duplex gap and is essentially adjacent to channel 37. For all practical purposes, the propagation characteristics of this spectrum are identical to those of the TV White Space. Using this spectrum, the wireless industry should be able to build systems that match or exceed the performance of systems in the TV White Space in every relevant dimension—range, speed, cost, whatever.
Was Interference Considered?
I could not find any mention of possible interference to systems operating in the TV White Space in the Microsoft white paper. But, unlicensed spectrum can be put to many uses. If the TV White Space performs as well as Microsoft describes, it would be a great band for surveillance cameras and telemetry equipment. Of course, if a car dealer across the street from a cell site serving a rural area were to install several surveillance cameras operating in the duplex gap, the nearby White Space cell site would not be able to receive signals from their users in the duplex gap.
Deployment of many short-range systems in the TV White Space would make the economics of providing Internet access in the TV White Space far worse than is projected in the Microsoft white paper. Certainly, any reasonable analysis of the cost and coverage of TV White Space service should consider the risk of such interference and should note that systems operating in the 70 MHz of licensed 600 MHz spectrum does not face a similar risk of unpredictable interference.
A second sign of progress in the debate is that the Microsoft white paper model looked at a mix of technologies and tried to identify the most cost-effective technology for each potential user rather than impose a single, one-size-fits-all solution. Microsoft recognizes that satellite-based Internet access is the least costly alternative in the lowest density areas. I could not determine whether the Microsoft analysis was based on the cost and capacity of the satellites in service today or of those that will be in service five years from now. Satellites are improving rapidly.
ViaSat 2 was launched last month. ViaSat 2 will have a throughput of 300 Gbit/sec—more than twice that of ViaSat 1 launched less than six years ago. ViaSat has started construction of ViaSat 3, which is expected to deliver about 1200 Gbit/sec but will cost about the same as ViaSat 2 to build and launch. (See Viasat’s 2016 annual report, page 30.) Over about a decade, the cost of satellite capacity will have fallen by more than a factor of six.
If Microsoft’s analysis is based on the capabilities and cost of the satellites operating today, then it is based on the performance of systems two generations behind those that will be in operation in 2022.
Consider a system with key parameters similar to those of ViaSat 3—capacity of 1200 Gbps and a cost of $700 million—call it SatX. Microsoft asserts that there are 24.3 million people in rural America that need affordable Internet access. If one assumes that there are 2.53 people in a household, then there are 9.6 million households needing such Internet access. Three SatX satellites would have a capacity of 375 kbps/household.
Of course, people don’t use their Internet access all the time or constantly at peak speeds. If one further assumes that the ratio of peak to average is 100, then a constellation of three SatX satellites could deliver a 37 Mbps (downlink) service to all 24.3 million people.
Of course, 9.6 million home terminals would be needed—those would cost a few billion more. $10 billion—the low end of Microsoft’s estimate of the cost of providing rural Internet access using the TV White Space—would buy three SatX satellites and leave $7.9 billion for home terminals, or about $800 per home. $15 billion, the high end of Microsoft’s estimate of the cost of providing rural Internet access using the TV white space alone would buy a five-satellite constellation delivering a 60 Mbps service and leave $1,200 per home to pay for terminals. An Excel spreadsheet showing the details of the above calculations and that allows one to analyze alternative assumptions is available here.
It appears to me that it is highly likely that next generation satellites will be the low-cost alternative for serving many or most of the 19 million that Microsoft identifies as best served by systems using the TV White Space. It also appears likely that the cost comparisons presented by Microsoft are based on last year’s satellite technology, not on the satellite technology that will be available in 2020 or 2022.
Microsoft supports keeping one TV channel vacant in every market. Doing so will have only a minor effect on the amount of TV White Space in many rural areas—even after repacking, Montana and South Dakota will have lots of White Space. Constraining the repacking to keep White Space available in the Northeast Corridor or Los Angeles will not improve rural Internet access.
Keeping a TV channel vacant in a location that would otherwise not have a vacant channel would increase the spectrum available at 600 MHz and below for such access services from 82 MHz to 88 MHz—an increase of less than 10%. (That 82 MHz is the combination of 70 MHz licensed and 12 MHz unlicensed.)
It is pleasing to see an advocate for unlicensed operation in the TV White Space recognize that such systems can serve only a small market. Similarly, recognition that satellites are the least costly solution for the most rural areas is refreshing.
However, the facts that Microsoft’s white paper (1) does not consider service provided by wireless carriers using 600 MHz spectrum; (2) fails to address the problems of interference among TV White Space users; and (3) appears not to have taken into account improvements in satellite technology, indicate to me that one should give relatively little weight to their conclusions about the costs of systems operating in the TV White Space relative to the costs of commercial wireless or satellite systems.
Similarly, Microsoft’s assertion that the FCC should force the upcoming repacking of TV stations to keep an additional White Space channel available everywhere—no matter the cost in impaired TV coverage—does not seem justified.
More detail, somewhat dated now, on this issue is available in two papers by Robyn, Jackson, and Bazelon: “Unlicensed Operations in the Lower Spectrum Bands: Why is No One Using the TV White Space and What Does That Mean for the FCCs Order on the 600 MHz Guard Bands?” and “Unlicensed Operations in the 600 MHz Guard Bands: Potential Impact of Interference on the Outcome of the Incentive Auction”.
Conflict of interest disclosure. Dr. Jackson has had clients in this area in the past. However, he has no current projects with any of those clients and wrote this on his own initiative without any prompting from such former clients.