dslreports logo
site
 
    All Forums Hot Topics Gallery
spc

spacer




how-to block ads


Search Topic:
uniqs
896
share rss forum feed

Guspaz
Guspaz
Premium,MVM
join:2001-11-05
Montreal, QC
kudos:23

Spectrum crunch: not a problem today, but eventually

There's a fixed amount of wireless spectrum available, and there's a theoretical maximum amount of data that can be pushed through it. That's a laws-of-physics type limitation, no new wireless technology will ever relieve us of that limitation. Is this a problem today, or even in the immediate future? No, there's a very long way to go before we hit there, and we can pump multiple gigabits of bandwidth over the air before we hit that limit. One day, however, we WILL hit that limitation. Cell sizes can shrink, spreading the load, although there are trade-offs there too. One day, in a few decades, we're going to run out of spectrum, and the aggregate wireless transit capacity at that point will be all we'll ever have.

So, the "looming wireless apocalypse" isn't a lie, it just isn't going to happen any time soon. But it'll really suck when it does, way down the road. Personally, I think that demand will eventually subside, making it less of a problem. After all, there are only so many things you can do with bandwidth. Video is driving usage ever higher and higher, but there comes a point where your video content exceeds human ability to perceive it (as in, your eyes can't resolve higher resolutions), and the ever increasing video bitrates taper off. We reached this point long ago for audio; you can transmit perceptually lossless 24-bit audio over a 3G connection without any issues, but nobody really bothers with that even on 24-bit capable hardware because it just doesn't make all that big of a difference.

I'd actually argue that, at the point where you can (or need to) stream perceptually lossless high-framerate 4K 3D video and surround sound audio, we'll have hit that point. Because at that point, you could just stream the entire interface to your device, and there would be no need for any higher amount of bandwidth to your device. That's probably somewhere on the order of one or two hundred megabits per second. Several orders of magnitude beyond current average usage (you might hit 20 megs on your LTE phone, but with a 2GB cap, your average ain't gonna be that high).
--
Developer: Tomato/MLPPP, Linux/MLPPP, etc »fixppp.org
Skippy25

join:2000-09-13
Hazelwood, MO

Re: Spectrum crunch: not a problem today, but eventually

Fixed or not, improvements or not, there is one thing for sure. There we NEVER be an exoflood as networks will always work themselves out in 1 of 3 ways or a combination of all 3.

1.) Carrier improves the network through investment in innovation and equipment.
2.) Carriers raise prices to cause people to leave the network freeing up resources for those that stay (simple law of demand).
3.) Carriers do nothing, people seek better alternatives thus freeing up resources for those that have stayed.

Nothing else needs to done period.
InvalidError

join:2008-02-03
kudos:5

Re: Spectrum crunch: not a problem today, but eventually

said by Skippy25:

Fixed or not, improvements or not, there is one thing for sure. There we NEVER be an exoflood as networks will always work themselves out

That depends on whether or not peak-hour usage will start tapering off before catching up with hardware capacity growth.

If you read Cisco VNI and other similar reports, some types of peak-hour usages such as video streaming grow by 50-60%/year. On the other hand, large routers's capacity/density/cost improve at a rate of about 25%/year.

Whether or not the exaflood will happen depends on whether or not demand will catch up with hardware's head-start before slowing down.

Alex J

@sunwave.com.br

Re: Spectrum crunch: not a problem today, but eventually

If you read Cisco VNI and other similar reports, some types of peak-hour usages such as video streaming grow by 50-60%/year.

Story above pretty clearly highlights Cisco's predictions are inflated. Two guesses why. One should be "to sell hardware."
InvalidError

join:2008-02-03
kudos:5

Re: Spectrum crunch: not a problem today, but eventually

said by Alex J :

Story above pretty clearly highlights Cisco's predictions are inflated. Two guesses why. One should be "to sell hardware."

Try looking around for companies that publish statistics about their usage growth that go into comparisons between average and peak traffic growth, they all agree with numbers similar to Cisco's... 30-35%/year average, 50-60% peak.

You do not need to believe Cisco, Sandvines, Akamai or anybody else, just visit the statistics pages from any internet exchange, they all show similar trends as well.

Here are the stats page from two of the largest internet exchanges in the world:
»www.ams-ix.net/statistics/
»www.hkix.net/hkix/stat/aggt/hkix···ate.html

That 50-60% does not only exist in Cisco's and other equipment manufacturers' papers, it also exists in the real world.
InvalidError

join:2008-02-03
kudos:5
said by Guspaz:

There's a fixed amount of wireless spectrum available, and there's a theoretical maximum amount of data that can be pushed through it. That's a laws-of-physics type limitation, no new wireless technology will ever relieve us of that limitation.

You are missing one parameter: fixed amount of bandwidth within a given cell. Reduce the cell radius by half, you can now reuse the same spectrum up to eight times as often (3D space) and you can repeat the process until you reach the smallest practical cell size such as pico-cells (50-200m range) embedded into incumbent modems/ONTs/CPEs turning every wired subscriber into a wireless cell for the incumbent's own wireless network.

For now, incumbents are sticking mainly to femto-cells (~10m range) due to the technical challenges of maintaining accurate frequencies over time but this will probably change once a cost-effective and reliable solution is found.

battleop

join:2005-09-28
00000

Re: Spectrum crunch: not a problem today, but eventually

That can be done but equipment, towers, and engineers are not cheap.
InvalidError

join:2008-02-03
kudos:5

Re: Spectrum crunch: not a problem today, but eventually

said by battleop:

That can be done but equipment, towers, and engineers are not cheap.

Most of the engineering is about picking locations, prepare mounting arrangements and getting uplink/power on-site. You have none of those problems with embedded pico-cells in subscriber CPEs since the wired subscriber provides all of the above for free... doesn't matter if individual pico-cell placement is sub-optimal if you have many times more cells than you need to handle demand. Worst case, you turn off some cells where there is excessive overlap.

88615298
Premium
join:2004-07-28
West Tenness
said by battleop:

That can be done but equipment, towers, and engineers are not cheap.

Us customers are in fact giving these companies tons of money. In the old days companies use to take profit and reinvest it to grow the company. Today that seems to be a dirty word.