Fred Golstein is one of my colleagues in the USA who is also assisting in writing the various reports we provide for governments and international institutions. His analysis below follows an international discussion on the technical decisions made by NBN Co and in particular the decision in relation to the LTE network. Several alternative technologies were discussed in that discussion and Fred mentions them below as well.
It strikes me that the Australian NBN choice of technologies to reach 100% of the population is quite well optimized. It stands in stark contrast to the US model, in which policies are based on nod-and-wink expressions of faith (that competition magically happens) and a heavy flow of subsidy cash (to incumbents).
The cost of wireline is largely one of linear mileage. If make-ready costs are kept under control aerial fiber on existing poles can be hung for something in the range of US$10-30k/mile. Incumbents have little make-ready cost, of course; they can overlash their old plant.
Underground construction costs vary hugely according to the type of ground. Direct burial or microtrenching into soft soil is competitive with aerial. New conduit in hard ground under streets (as we have in New England) is more like $500k/mile. However pulling fiber in existing conduit is cheap – then it all comes down to who owns it.
But once you get to rural areas linear mileage adds up. So NBN’s plan to use LTE for the 93-97th percentile seems ideal. Wireless costs are generally a matter of square mileage. While WISP radios are cheaper than LTE, they have a shorter range and less capacity. They are optimal for small competitors, small areas (like urban competitive providers), or very low density areas. But an LTE radio access node, even at $100k or more, can cover a much wider area with its stronger signal. If it can pick up 100 customers in a 10-mile radius then it pays off. Fixed LTE has a wider range than mobile, since the antennas are outdoors and can be directional.
When you get to the outback, only satellite makes sense. Terrestrial backhaul alone would be prohibitive. Newer satellites have more capacity than old ones, but the latency will always be an issue. Such is the cost of distance.
DAS (Distributed Antenna System) doesn’t really fit the fixed equation. DAS is a system in which the front end of a cell and antenna are on the pole – a group rather than one single antenna element – with the rest being backhauled over fiber. It’s a good way to get very high urban capacity out of a wireless mobile network, but it still requires fiber to each radio node, generally on each block. This is how ATT Mobility can raise its capacity without buying T-Mobile and ignoring the traffic its subscribers already generate.
‘Mesh’ refers to many things. The original WiFi meshes were cute but almost useless. One radio for access and meshing leads to congestion (see Ricochet). Multi-radio meshes, which come in different flavors, tend to work better. This is a nice way to do fixed wireless on the cheap, especially for WISPs with no spectrum licenses. But LTE, with its higher power, will outperform it. Since NBN is a government creation it no doubt has ample spectrum to work with.
White Space technology is a different story – about making more efficient use of spectrum that is nominally reserved in order to protect something that doesn’t need that much protection.
Fred Goldstein http://www.ionary.com/
We invite your comments: Please click here to commentTagged in: Australia, National Broadband Network, NBN, NBN Co