 Wireless backhaul works fine for a good majority of 4G cellsites and fiber isn't necessary for most, I've learned this year. The "received wisdom" was that you needed fiber for the 30-150 megabit speeds 4G towers required. Some other things I've learned lately that effect broadband buildouts include:
- While commercial interests aren't building out towers everywhere, they are putting transmitters almost everywhere a tower exists. Verizon wants to put a radio on virtually any available tower with a reasonable price (including backhaul,) I hear from the field. It's their strategy to win customers by having the best coverage, and the other carriers don't want to be far behind. Verizon's announced plans for 92% LTE by 2013 confirm that, especially because they are already discussing how to go even further. On the other hand, I've found no evidence of a major commercial buildout of towers into the areas not covered. That suggests a strategy of focusing subsidies on building towers + backhaul while expecting the carriers will add the transmitters and service.
- While getting more spectrum will be a good thing, increasing the efficiency of spectrum use has a payoff that will be even larger. I'd guess some practical efforts can effectively raise efficiency in the available spectrum 5-10 times, although I don't have solid data behind to support that yet. <hr id="system-readmore" /> AT&T is ready to distribute 5-10M femtocells that will effectively almost double their capacity. Adelstein's "Use It or Lose It" rules on license renewal would effectively bring into use 10's of megahertz - or more - in urban areas and potentially hundreds in extreme rural areas. Without a penny of public money, wireless coverage can easily expanded to 98-99%. It's ridiculous that the carriers don't share buildouts, as Cingular and AT&T Wireless did. Bellsouth/Cingular VP Ron Dykes pointed out the cooperative build saved "hundreds of millions" in New York alone, and would also allow more intense use of the spectrum as the different carriers had different peak use patterns. Extending the "white space" rules into the remainder of the spectrum is natural; there's no reason a radar system near the Arctic needs to prohibit use of the same spectrum in Florida. Ultimately, except for the most crucial needs, we need to change the rules from "no interference ever" to "minimal interference," allowing cognitive radio and related technologies.
Some economists believe allowing easy sale of spectrum will have much of the same effect; it's possibly a good thing, if at least part of the windfall profits are taxed, but inefficiencies in the market mean we need to go much further.
- New generation satellites, the first of which is scheduled to launch in 2011, will allow speeds of 5 to 10 megabits to many. Viasat CEO Mark Dankberg surprised the FCC broadband hearings by telling us the latency would also be significantly reduced. They can't do anything about the speed of light across 22,000 miles, but they can relieve congestion on the satellite itself which has been part of the problem. Caching and some tweaks to the protocols - reducing acknowledgment packets, for example - also can speed things up. It will be years before the last 1-2% of homes have a choice other than satellite, so improved satellite is a natural part of the plan.
- The congestion problem on the cable local loop can be dramatically reduced and probably almost eliminated. When the FCC criticized Comcast for blocking Bittorrent, their engineers went to work on improving the network. Within months, the problem of upstream congestion on Comcast nearly disappeared and they were able to turn off the Sandvines for the vast bulk of the time. That's partly because p2p traffic is down dramatically as a percent of traffic and possibly absolutely. More interesting is that a modest increase in capacity was enough to solve nearly all the problem. The cost was so low that it was invisible in their capex budget. In retrospect, the problem was that the cablecos deferred ordinary upgrades while waiting for DOCSIS 3.0. With 3.0 upstream now shipping, problems on well maintained cable networks should virtually disappear for years. Comcast's Tony Werner tells me that upstream will deploy widely starting in 2010. My guess is that 40-70% of the U.S. will be covered with 120 shared/?50 meg typical upstream speeds in 2011-2012, but everyone is waiting for results from the field before making firm plans.
- Somewhere between 2M and 4M homes get cable TV but not data, most of which can be upgraded for less than $500. 1M homes upgraded for $500M - some in loans - would reduce the "unserved" by 15-30% for about 8% of the stimulus spending. That requires solving the high cost backhaul problem as well, but "special access" is high on the FCC agenda for the areas where the backhaul carrier is gouging.
- The big cablecos have half a million "unserved" homes, perhaps 10% of the "unserved." Most have over 99% coverage, but even a fraction of 1% adds up. Comcast has refused to answer three requests I've sent over for information about where these "unserved" are; presumably they will tell the FCC one day since you can't make a broadband plan without information like that. The feds put $7B on the table to help reach these folks - none of the MSOs will even recognize the problem, or take government money to solve it.
- DSL bonding doubles effective speed at modest cost (<$100 and soon less than $50.) Bonding has been in the standard for years, but only this year have cost effective equipment become available and proven in the field. With line counts down, there's plenty of copper available. When the telcos get their deployment organized, 30M homes can double their speeds.
- Dynamic Spectrum Management - an inexpensive technique for reducing noise - has now been deployed to 20M lines around the world. Besides increasing speed, it dramatically improves reliability of longer loops. This allows connecting many now turned down by the carriers. Bonding and DSM mean service can reach a significant percentage of the currently unserved by DSL. I've no way without the company data to estimate closely, but up to a quarter of those not offered service in 2008 will now be servable.
- DSL repeaters now go to 6 megabits downloads. Many homes between 15,000 and 25,000 feet can now be served at megabit speeds. The repeaters cost less than $200, and can probably reach a third or more of those currently unserved. Because they can be deployed on single lines, they are the right technology for the 1 3, or 5 homes that are not part of a larger cluster.
- 90% of Clearwire's 4G cell sites are connected wirelessly, saving the cost of running fiber. It turns out that line of sight is usually available, if necessary by putting up one very high tower visible to the others. I've reported that reduces backhaul costs in most places to $8,000-25,000
I jotted down an outline of these thoughts on the train to D.C. to meet with some of the folks working on the Broadband Plan. I didn't do a "presentation" but in the spirit of disclosure I wanted to review them publicly. I wrote this quickly in that spirit. If I have any errors, please let me know ASAP to correct them. These are my comments, not theirs. On October 16, I had the chance to talk with Blair Levin, lead economist Scott Wallsten, and several staffers. |
