The Google Theory of Spectrum
Wednesday, November 21, 2007Below is an exchange between Spencer Michaels of the Lehrer Report (shown on November 20 on PBS) and Eric Schmidt, CEO of Google and Andy Rubin, director of mobile platforms. I have included it in this blog because I was also interviewed on the program and while they shot about an hour of video, only a few minutes was used. Spencer asked a question about wireless bandwidth and when I was taped, I had no idea that the question would be asked of the Google folks, but I have to admit that I am astonished at the answer given by Eric Schmidt and Andy Rubin.
SPENCER MICHELS: I talked to an industry analyst a week or so ago, Andrew Seybold, and this is what he said and I'm kind of curious as to how you would answer it. "Google's view of the wireless world is an extension of the Internet. And as much as I respect Google, the wireless industry can't be an extension of the Internet because wireless bandwidth is finite. It's a fixed resource and a shared bandwidth. The more people who use it in a given area the less data speed they have, so you can't take the Internet model and just move it to the wireless world. You have to change that model a little bit as you move forward." Is that fair criticism?
ERIC SCHMIDT: That's exactly the same criticism that was said about the wired Internet 15 years ago. That somehow the wired Internet would not scale, would not grow, that we would not learn how to build applications that could work for example in shared what are called hybrid fiber coax networks which you have at your home. The fact of the matter is that the industry can solve these problems and solves them very well. I completely disagree of the characterization that somehow the wireless network is going to be any different than the wired network. People want to use the Internet and they want to use it at home and in their office and when they're on the go and when they're on the airplanes. And they want to use the same powerful applications whether it's a personal device that they're carrying or on their desktop or at the beach.
SPENCER MICHELS: People want to use all the water they want to use, but there isn't enough water, so some people don't have enough water to water their lawns. That doesn't mean it's there.
ERIC SCHMIDT: Technology is different and the technology can create things out of nothing. The fact of the matter is that there's enormous spectrum becoming available through licensing programs, better radio design, faster computers, and so forth that in the next five or 10 years most of us will be carrying around devices that could speak on networks that are faster than the networks we currently use.
SPENCER MICHELS: Do you agree with the boss?
ANDY RUBIN: Absolutely. This is Moore's Law for wireless, or Moore's Law for spectrum. You know when we're on the wired Internet, in my day we had dial-up modems. Then we got DSL, then we got broadband, we're in the megabit speeds. Same thing happened to wireless. Remember those old analog phones where you'd get a dropped call every mile you traveled? Now we have digital phones, we have newer modulations that are helping us pack, densely pack, more calls in a single channel. And I'm just very optimistic that we'll have newer modulations throughout time that'll just get us faster and faster in the broadband areas.
SPENCER MICHELS: So Seybold saying that the bandwidth is finite, is it relevant?
ANDY RUBIN: Technology doesn't stand still.
I sent this off to another senior person at Google and got back a very well written email explaining why Eric answered the way he did. I did not ask for permission to use his comments here so I will paraphrase. Basically, he sees networks growing to accommodate the traffic, cells being built out by the millions, cell sites for existing networks being closer together, the use of micro, pico and femto cells, and networks such as Wi-Fi that will be available at shoeshine stands, gas pumps, subways and every hundred seats at a stadium.
In other words, as the amount of wireless data usage grows, networks will grow to keep up. He pointed out that during the short period of time that high-bandwidth applications such as Skype/VoIP, Earth, Maps, YouTube, etc. grew in number and popularity (about a year), Internet bandwidth increased. The bottom line from Google is that “the Mobiles as the new PC era is not facing a bandwidth wall.”
I wish I shared that optimistic view of wireless. However, having been in the trenches and worked with networks and new technologies for more than thirty years, I tend to be more of a realist (I guess Google would call this unenlightened). Yes, there are technology advances and yes, you can build cells close together and add more capacity on demand, but for each cell you deploy you must find a way to backhaul the voice and data from that cell. Mesh networks are one way to cut down on the number of backhaul links and being on a fiber ring is another, but I don’t believe we will be able to break Shannon’s Law. The goal of driving down the cost of wireless while making it ubiquitous is not realistic or even achievable while maintaining a cost structure that supports the necessary investments.
My other concern about this vision is how reliable this combination wide-area and local-area “network” will be. Will it be sufficiently robust to withstand power outages, high winds, floods, earthquakes and fire? Wireless network operators have been working diligently to make their systems robust and have made major investments to accomplish this. With each major disaster, we find fewer and fewer failures. But we also see sharp increases in peak usage of commercial networks by first responder agencies who use them to off-load their emergency communications networks and by the general public trying to reach their loved ones and friends.
What happens when the power goes out and stays out for hours? Will this combination network continue to function?
Google might be able to push technologies and build out a zillion access points to be used in conjunction with the wide-area networks, but how will it keep this type of network up and running in critical times? Are we going to trade “unlimited” bandwidth for reliability? I, for one, hope not!