article in today’s Toronto Star describes an interesting innovation
of Sault Ste. Marie, ON’s local power utility company to provide
Internet access over the power grid.
While this technology has been put to use in Europe, it has always been
considered impractical in North America because North American electric
systems step down to household voltage at the pole, necessitating a
local transformer through which the network signal is destroyed. In
Europe, the stepdown to household voltage is far enough away that it
is practical to locate the off-grid network uplink after the
The Sault Ste. Marie solution is exactly the opposite: take it off
the grid immediately before the transformer with a wireless
access point. Data between the WAP and substations travel over the
power lines, and a fibre link at the substation connects that end
of the connection to traditional networking equipment.
The article suggests that this is advantageous because power companies
have a last-mile advantage over phone and cable in some areas, but my
inclination is exactly the opposite: the “last mile” here is the 150m
range of the WAP. Phone and cable companies already have pairs going
into houses that they can share for services, so there’s really no
last-mile infrastructure costs to provisioning DSL or cable internet.
But a 150m range means that the power company is going to need a
WAP every 250m or so, and that sort of equipment outlay sounds a
lot like laying a brand-new last mile to me. And Sault Ste. Marie
is a small city: 74,000 people in 2001 with a 6000-person drop between
1996 and then. Pulling a 10m average frontage out of thin air,
that means one WAP can probably serve about 50 dwellings (with overlap
to the next WAP), which means you’re looking at a couple thousand WAPs
to serve the city, and that the MTBF of the wifi last-mile is 1/1000
or less of the MTBF of a single WAP. It appears you can find outdoor
WAPs with 20kh-30kh MTBF, which means you’d be rolling a truck to
replace a WAP every two days. Somehow that doesn’t seem to scale to
populations over one million.
On top of that, you’ve got a related business problem: neighbourhoods
with low adoption rates cost significantly more to network. With
DSL, the hardware deployment centres around the CO, and you can
buy sufficient DSLAMs to meet projected need, wherever that need
might come from — but one customer in a neighbourhood means
that that neighbourhood needs a WAP.
It’s certainly a neat way of solving the problem of grid internet with
transformers on the pole, but I can’t see where it makes much business
sense — but I admit that it would be pretty neat to have your laptop
just work anywhere in town that has power and a nearby WAP.