Fast Net News, DSL Prime, DOCSIS Report

Bengt Nordstrom and friends drove around Stockholm and blogged LTE downloads "above 25 Mbps more often than below, and we’ve reached 45 Mpbs downlink on some occasions." Service frequently dropped, confirming the rumors I've been hearing of profound issues still unsolved.

   He first blogged he was only getting 12 meg down, which made news around the world because TeliaSonera is claiming they were offering "up to 50 meg." Nordstrom is a wireless analyst with decades of experience. He's clear this wasn't scientific testing, but it's interesting and better than many expected.

     Wireless is shared, meaning the speed will go down as the network gets loaded. So I'm still comfortable with Verizon CTO Dick Lynch's estimate of speeds from 5-12 megabits in regular use. Until proven otherwise, I speak of LTE speeds as "megabits", not tens of megabits. In addition, the guesses are that heavily used LTE should be sold mostly as a 3-7 megabit service. In particular, if widely used to watch quality video, the reliable speeds will be far below the "up to" figures. Nearly all the carriers are considering a very low cap, too low to support video watching for more than a few hours/week.

      Verizon and just about everyone else thinks that LTE is a niche data play until 2012-2013, when voice and data phones become plausibly priced. Verizon has a huge buildout for 2011-2012, but their predictions of customer numbers are low until sky-rocketing in 2012-2013. The chipmakers are (privately) making similar forecasts, with the necessary low-power highly integrated chips promised for 2011 but volumes probably 2012.

     Even at 3-7 meg, LTE is certainly enough to take customers away from the low end of DSL and cable. That's starting to happen even with 3G speeds, one reason DSL/cable net adds are dismal many places. Plenty of people want to check there Facebook and email and occasionally surf, not do volumes of video and other apps that need bandwidth.

   Since most of the developed world is quickly going to 50 megabits (90% of the U.S., half of France, Britain and Germany ...) wireless will be only a partial substitute, only rarely a replacement for DSL or cable. The early results in the broadband delphi had estimates of 5-25% of homes going wireless only by 2015. If Verizon or AT&T chose to offer LTE at a DSL price - plausible out of territory - the cord-cutting could be huge. But that's not likely - the duopoly is more comfortable at higher prices.

    One of the world's most respected network engineers tells me to think of 1.8 megabits/megahertz as an excellent result in the real world. The vendors are promising much more, of course. In a typical LTE deployment of 20 megahertz, that's 35-40 megabits shared. With a lightly loaded network, Bengt's 15-30 megabits makes sense. That will be difficult or impossible to maintain in a heavily loaded network.

    That opens interesting possibilities in rural areas, predictably a light load. It would almost certainly require larger spectrum blocks, but most of the spectrum in rural areas isn't used today. All the carriers sharing the spectrum and towers would be the natural build, and two FCC technologists suggested it at the workshops. The carriers instantly said no way would they accept changes, and in D.C. at least the carriers win 90+% of the time.