Tales of the Sausage Factory: Open Spectrum Doubters

A mailing list I’m on pointed to this rant by Chris Davies against open spectrum, and asked for a response after it was cited approvingly (if confusingly) by Corante. While I am tempted to respond simply by reference to the filksong by Brenda Sutton, I will attempt a more substantive answer below (although everyone really should buy Rite the First Time to hear that song and others).

Chris Davies takes on Kevin Werebach’s open spectrum white paper, in which Kevin basically proposes that we can over time do away with exclusive licensing and let the technology handle intereference problems. As regular readers know, I am a big believer in the power of unlicensed spectrum, which can beat the snot out of the power of cheese with one arm tied behind its back. So I naturally was curious to see where Kevin made his fatal error.

As one might imagine, I find more lacking in Chris’ arguments than Kevin’s. In no small part this is because Chris reads a lot that isn’t there into Kevin’s long range utopian projections. But it also comes from a basic skepticism about the existing tech which seems somewhat unjustified to me.

Chris first starts with the “crowded room” analogy and demolishes it. As it happens, I am also not a big fan of analogies. For those unfamiliar with this one, people like to comapre radio speech to audio speech and observe that just as lots of people can chat in a crowded room, lots of radios can communicate without interfering with a higher-powered licensed service.

Chris raises some problems with this analogy, but that doesn’t disprove the case against unlicensed spectrum. It simply proves the limitations of arguing by analogy. I’m down with that. Drilling a bit further, the objections appear to be:

1) We don’t really have a good answer yet to frequency hopping and can’t put the intelligence in devices that open spectrum advocates believe we can;

2) We never will be able to do so, and the proposed solutions (like mesh) don’t work either;

3) Quality of Service (QoS) would suck rocks in open spectrum;

4) If you throw in malicious people deliberately trying to screw up the networks, the problems become insoluble.

5) Unlicensed isn’t really innovative anyway, since it all rides on wi-fi.

This all seems horribly reminiscent of arguments I heard in 1995 about whether the Internet was really going to catch on or really going to change things. On the one hand were techno-utopians who believed that TCP/IP was the pathway to freedom and economic empowerment and nothing could stop it. On the other were those who pointed to very real technological questions- would people really invest in services with no QoS and only promised “best efforts” to deliver traffic? Would routing tables fall apart because they couldn’t scale? Could volunteers and (relatively) low paid government contractors really provide the basis for an international telecommunications medium? What about congestion at the network access points? Would the private sector really build a backbone once NSF privatized NSFNET?

Well, ten years later we know how things turned out. The technical objections were real, but there were technical solutions and a very good economic case driving adoption.

I see the same factors coming together for unlicensed spectrum. Certainly there are technical problems, but they are not insoluble. There is no physical law preventing the open spectrum model — radio waves do not collide with each other or anything like that. We can see sufficient proof of concept of potential answers in architecture of networks and the technology of devices to give confidence (to me anyway)that we will continue to find solutions to these challenges.

And the economic case is there to drive deployment, because the cost of these devices is cheap when compared to the wireline and licensed spectrum alternatives. There are a lot of people happy to by less than perfect QoS service when they are cheap, just as there are plenty of people who insist on high QoS because they need it. That’s why a free market where anyone can offer whatever services they want is useful, because it gives people choices.

The biggest problem right now, it seems to me, is shortage of usable spectrum to give this stuff room to grow. That’s whay I’m doing what I do at the FCC. If the FCC authorizes the 3650-3700 MHz rulemaking, then we will see what happens in a band when you have relatively high power for point-to-point links.

Which brings up another rather serious criticism of Chris’ critique. He does not factor network architecture as a means of addressing the interference issues with relatively high power devices. He also is skeptical of the ability of radios to ignore undesirable same channel or near channel signals. While I hate to use bad analogies again, his “person bellowing” and thus drowning out his neighbors argument put me in mind of an item on NPR last night about a class at Juliard where students learn to perform under pressure. The teacher trains the students how to focus and use their nervous energies to ignore distractions, including loud noises going off at random right next to them.

As always, an imperfect analogy. But we can already teach devices to recognize set words and voices. The problem of recognizing only a desired transmiter is harder, but not impossible, even without resort to mesh. (Like Chris, I’m not a big believer in mandatory mesh where, to join a network one must become a node, but there are lots of mesh networks that do not behave in this fashion.)

My biggest criticism of Chris Davies article is his knock at the end on whether unlicensed is “innovative” because it relies entirely on “wifi”. To me, this is like saying the internet is not innovative because it all centers on TCP/IP. Sure, the 802.11 family of protocols provides the means by which all these devices communicate. It’s what these devices allow you to do and the inventive and economic activity they stimulate that is “innovative.” The Consumer Electronic Show (CES) is wall-to-wall gadgets this year that work to integrate various home devices wirelessly. Yes, they all rely on 802.11 “wifi.” That shared platform allows all the other activity to take place. Nor is it just consumer devices. Folks like CUWIN can deploy their mesh network software that enables communities to roll their own networks. Sure, it’s just “wifi” and open source, but figuring out how to turn old computers and cheap radios into neighborhood-wide connectivity strikes me as pretty damn innovative.

To close with a new verse to the afore cited filksong:

He said: “someday we won’t need a license from the FCC

To send each other pictures of our children wirelessly

And it’ll be so cheap to build ’em that the networks will be free.”

The crowd said “Kevin Werebach you are crazy!”

Absolutely bonkers, bouncing off those padded walls,

Completely crackers, looney tunes, buh-duh buh-duh buh-duh–that’s all.

The lights are on, nobody’s home, moronic vegetable,

Bet yer bottom dollar, you are crazy!