[LINK] tRe: NBN white-elephant-to-be ...
rchirgwin at ozemail.com.au
Thu Aug 26 10:10:23 AEST 2010
> On 26/08/2010, at 9:34 AM, Richard Chirgwin wrote:
>> Let's pick up cells of max size 500 metres, and ignore two side issues
>> (how much will 100,000 towers in Sydney cost, and will people tolerate
>> wireless towers everywhere?). Let's just stick to performance.
>> From page 399 of "3G Evolution: HSPA and LTE for Mobile Broadband",
>> (Dahlman, Parkvall, Skold and Beming, Academic Press, ISBN
>> 978-0-12-372533-2), there is a graph charting LTE capacity.
>> If the system is using 5 MHz radio channels, the performance at 500
>> metres from the base station is 20 Mbps; with 10 users, the per-user
>> budget is about 2 Mbps.
>> The charts for performance with 20 MHz radio channels are harder to
>> interpret, but the best total throughput of the entire radio channel is
>> about 110 Mbps - about 10 Mbps if there's ten users in the cell (my
>> interpretation of the chart given on page 404 of the same book).
>> So: let's assume that whether we're in the city or the country, we can
>> serve almost adequate bandwidth to ten users in one cell; that the total
>> capex of a base station (with three cells) is about $90,000. We get
>> 90/30 or $3,000 capex per household. It's not so much cheaper than
>> pulling fibre.
> Richard, your figures are based on 10 active users per cell. But in a 500-metre radius of a tower, in suburban or outer metropolitan areas, surely there would be many more than 10 households in a 500-metre radius of the tower? How does this affect your back-of-envelope model?
> Also, if we're talking about wireless as ubiquitous last-mile connectivity, and you can only provide acceptable speeds to 10 simultaneous users per cell, then presumably at peak household broadband times in the evenings that means a cell per every 10 households not for every 10 simultaneous users. Or maybe every 20 households, given non-users and people not at home for the evening and so out.
> That brings us back to Narelle Clark's comment, "In order to get those 100 megabit speeds and beyond you'd need to be installing a base station around about on every suburban block," she said. "At the end of every street there'd need to be a base station."
> I'm looking at a satellite image of Yagoona in Sydney's west, as random an Australian suburb as any with its streets of red-tiles houses. Counting 10 houses, it does indeed look more like a base station every 100m, not every 500m.
> Disclaimer: Due to Bavaria, this morning my arithmetic may not be reliable. Caveat lector.
All valid and correct objections, Stil. I was trying to be generous, and
frankly the LTE book makes my brain hurt, so I was lazy as well as
generous with my maths.
More information about the Link