Last Wednesday (April 18th) I testified at a hearing of the House Technology and Innovation Subcommittee on the spectrum crunch, along with a panel of well-informed people (see webcast here.) When you testify at a Congressional hearing, you prepare some written testimony, you deliver an opening statement, and then the questions start. More than anything, the committee wanted to know if the spectrum crunch is real, and if so, what the consequences will be for innovation and what the government can do to help.
The notion of a capacity crunch in any kind of network scenario is actually a bit subtle. We use networks indirectly, as means to run applications, to get content, and to use services. When network capacity is constrained (which it always is in one way or another) we use the applications that work well, and have little awareness of things we might be doing if capacity were greater. Why would app developers pour resources into developing apps that nobody can use? Developers want to make money and have lots of people using their code, so they simply don’t waste their time on things that are utterly impractical.
So the first challenge is to point to things on the horizon that aren’t quite practical yet, but which may be if capacity grows. In the mobile space, the near horizon is occupied by apps such as Mobile Augmented Reality that allow mobile users to interact with cloud services through video streams, location, and social networks (see the testimony for details.) MAR isn’t just spectrum-hungry, it’s also very resource-intensive on the processing and storage side, and revenue models need to be developed to pay for all those resources, of course.
People with a taste for advanced mobile apps have been experiencing the crunch in particular times and places since the 2G iPhone was introduced in 2007. The crunch was especially severe in San Francisco, where it’s hard for carriers to get zoning approval for new towers, and this has led to some high-profile breakups on the part of SF bloggers and their phones.
If the projections made in the National Broadband Plan are correct, by this time next year smartphone users will begin to experience annoying numbers of dropped calls, slow access to web sites and app services, and we can pretty well forget about augmented reality. Having established that, we then went to discuss practical steps to bring more data capacity to mobile apps.
The situation is especially dire because the actions that the government’s lead spectrum agencies (FCC for commercial spectrum and NTIA for government spectrum) have attempted to make have been rebuffed by Congress and uncooperative agencies such as the Defense Department. Congress prevented the FCC from repacking TV channels or otherwise pressuring TV broadcasters to take part in the incentive auction. They also allocated the D block to public safety instead of auctioning it and using the proceeds to pay for public safety networking on the commercial networks. And they raised a series of political questions about the FCC’s conduct in the LightSquared case that are wholly unwarranted.
The net result of these actions is that we have more spectrum in government hands than we did when the plan was issued, not less. So we’re literally moving backward as a nation toward the 500 MHz goal. This is a real problem, and it needs to be addressed in a serious way. This won’t happen until we get some serious media attention, however.
The day of the hearing, the New York Times published an article by Brian X. Chen titled “Carriers Warn of Crisis in Mobile Spectrum” that cast a shadow over the hearing. The article was so full of spin I’d have guessed it was written by a flack working for a PR firm if I didn’t know better. The title is a dead giveaway: The notion of a “spectrum crunch” isn’t something that the carriers made up, it was in the National Broadband Plan. Why attribute a notion to a commercial interest that was part of a highly respected government plan?
The article asks a question that’s meant to be provocative:
But is there really a crisis? Some scientists and engineers say the companies are playing a game that is more about protecting their businesses from competitors.
That’s right, not only is the writer questioning the broadband plan’s forecasts, he’s offering up a cheap little conspiracy theory in its place. Still, raising the question as to the validity of the forecast doesn’t doom the article to sensationalism; that comes from the selection of “scientists and engineers” with various axes to grind to tell the writer’s story.
Chen’s technical sources are Marty Cooper, David Reed, and David Isenberg, three people widely known in the networking world for their contrarian viewpoints. Cooper was an engineering manager at Motorola when they did their pioneering work on car phones and the early mobile handsets. Cooper is one of eight people cited in Motorola’s first cell phone patent, a very important piece of work after the Bell Labs patent that was the real ground-breaker on cellular technology. Chen calls Cooper the “inventor of the mobile phone,” an enormous stretch since mobile phones date back to 1880, long before Cooper was born. It’s not unreasonable to say that Cooper was one of the inventors of the cellular phone, but not that he was the sole inventor or even the principal one. Engineering managers frequently put their names on patents created by engineers who report to them, but managing an inventor is not the same thing as being one.
Twenty years ago, Cooper co-founded ArrayComm, a company that was supposed to have a solution for the spectrum crunch, the smart antenna. While that term rolls off the tongue as easily as smart phone, the company has yet to make good on the promise. They’ve given some impressive demonstrations, but seem to make most of their money from patent licensing and software, not from smart antennas.
Cooper’s claim is that wireless technology doubles in capacity every 30 months. Chen doesn’t seem to realize this is “Cooper’s Law,” a claim that Cooper has made for many years. Its importance is that radio technology advances much more slowly than computers do; the corresponding law for semiconductors is “Moore’s Law,” the claim that semiconductors double in performance every 18 months (and double in density every two years.) Placing the two laws side by side doesn’t provide confidence that advances in radio technology will ever keep up with the demands that come from the chips that power smartphones. Chen doesn’t notice this either, and he doesn’t notice that 20 years of ArrayComm hasn’t solved the mobile capacity crunch.
Cooper’s remarks are reasonable compared with what’s to come, however:
Arguing that the nation could run out of spectrum is like saying it was going to run out of a color, says David P. Reed, one of the original architects of the Internet and a former professor of computer science and engineering at the Massachusetts Institute of Technology. He says electromagnetic spectrum is not finite.
Mr. Reed, who is now senior vice president at SAP Labs, a company that provides business software, explained that there are in fact newer technologies for transmitting and receiving signals so that they do not interfere with one another. That means separating the frequency bands would not be required — in other words, everybody could share spectrum and not run out.
The reason spectrum is treated as though it were finite is because it is still divided by frequencies — an outdated understanding of how radio technology works, he said. “I hate to even use the word ‘spectrum,’” he said. “It’s a 1920s understanding of how radio communications work.”
First, Reed is a bright guy, but he’s not an “original architect of the Internet.” The Internet Society has a Hall of Fame for the original architects, people like Pouzin, Davies, Roberts, Kahn, and Cerf, and Reed’s not among them. Reed was a grad student in the mid-70s when the final polishing was done on TCP/IP, but all the important decisions had been made before he came along. He’s sometimes called the “creator of UDP,” but that’s an Internet in-joke because UDP is essentially an empty placeholder in the Internet’s protocol family. He acknowledges that himself and doesn’t make such grandiose claims. And even if Reed were the one and only father of the Internet, that status wouldn’t endow him with any particular insight regarding mobile networks.
More disturbing is the way Chen interprets Reed’s remarks about coding systems. What Reed would have told the writer is that it’s now possible to build radios in such a way that multiple transmitters can piggyback on the same radio frequency in the same cell. They can in fact do this in such a way that their data streams are distinguishable by a smart receiver. Such technology, Code Division Multiple Access (CDMA,) is already a part of every 3G and 4G cellphone.
There’s a limit to how much of this sharing is possible, and there’s also a loss of per-transmitter performance when it’s in use. Additional developments such as Space Division Multiple Access (SDMA) and Multi-User Multiple Input and Multiple Output (MU-MIMO) relax some of the constraints on CDMA, but they don’t eliminate them. Reed is describing a hoped-for development in radio receivers that doesn’t actually exist yet and probably won’t ever fully exist because of information theory constraints. While it’s possible for radio receivers to share frequencies with their neighbors, there are always going to be limits on the amount of information that can be moved through a given frequency, regardless of the number of transmitters.
The argument that “spectrum is infinite” is Zeno’s Paradox of Achilles and the Tortoise; it’s an infinite regress that divides a range into an infinite number of smaller ranges. It’s a logical fallacy that presupposes distance, time, or frequency to be both finite and infinite at the same time. Pulling the wool over the poor reporter’s eyes with this nonsense is very poor manners, but connecting it to a carrier conspiracy to extort high prices from iPhone users is enough to induce Chen to drop his critical filter. Reed claims that every carrier is a monopoly, which is torturing the English language.
Chen’s other expert is David Isenberg, an open Internet advocate who worked on the National Broadband Plan but who, like Reed, has no background in wireless networking. Isenberg accuses carriers of engaging in capitalism:
David S. Isenberg, who worked at AT&T Labs Research for 12 years before leaving to start an independent consulting firm, said the carriers have been deliberately slow with adopting more advanced radio technologies. He said that spectrum licenses come with obligations where carriers had to agree to serve the public interest, but those agreements have significantly weakened. “Their primary interest is not necessarily in making spectrum available, or in making wireless performance better,” he said. “They want to make money.”
If Cooper, Reed, and Isenberg were all correct, doesn’t it seem likely that an upstart carrier such as T-Mobile, MetroPCS, or Leap would realize that there’s also money to be made by adopting these “more advanced radio technologies” in order to reduce the restrictions on their services? They could buy base station software from ArrayComm, put some smart antennas in an Android phone and sell a service with infinite performance in any allocation of spectrum.
Any such company would rule the market for smart phone service in short order. They wouldn’t have to depend on chips built by companies such as Intel, Qualcomm, or Broadcom (who are presumably in on the conspiracy,) they could design their own parts and have them made by a contract fab, the way most communications chip companies do. There are plenty of contract manufacturing firms in Taiwan and mainland China who would be happy to put all the parts together and even certify them with the FCC.
The reason this hasn’t happened is that it’s simply not possible with the technology we have today. No matter how much signal processing you can do in a mobile phone, there are real limits to the number of bits per second that a radio frequency can carry. Some of these limits are directly related to battery life, and some are related to the ability of digital and analog systems to work together to impress information on a radio wave. The communications industry has made steady progress toward improving radio efficiency for a hundred years, as Cooper points out, but the constraints are very real.
So how does a fanciful conspiracy and science fiction piece like this make its way into the nation’s most influential news organ the day of a House Technology and Innovation Subcommittee hearing on spectrum innovation? I can only guess, but it appears that the New York Times is no longer content with its traditional role of reporting “all the news that’s fit to print” and now seeks to become more blog-like. The article was put into the record by Science Committee chairman Ralph Hall (R-TX) and was a big part of the buzz around Washington all week. This is sort of thing we’ve come to expect from TechCrunch:
But with Mike’s departure, the gonzo spirit that first drew me to TechCrunch — that desire to not just report the story, but to be part of it — has gone.
[emphasis added] If this sort of thing continues, the line between responsible journalism and sensationalistic, link-happy, fear-mongering blogging will be erased, and that would be an even bigger problem for public policy than the very real spectrum crunch. Let’s hope it doesn’t.