dslreports logo
 
    All Forums Hot Topics Gallery
spc

spacer

Search Topic:
uniqs
2763
share rss forum feed

bernardc

join:2009-12-13
Berkeley, CA

[Speed] Download/upload ratio: Is it physics?

This is a question for DrDrew or someone of his caliber, if there are any here: In modern cable modem systems, the marketed download speed is generally several times the upload speed. Is this because of the physics of the wire system, i.e. asymmetrical SNRs? Or because of the costs at the head end and beyond? Cost of the modem at the customer end? Or what?

Thank you, DrDrew for all of your fine work here.


tshirt
Premium
join:2004-07-11
Snohomish, WA
kudos:5
DOCSIS is asymmetrical by design, which wikipedia can easily explain
»en.wikipedia.org/wiki/DOCSIS

bernardc

join:2009-12-13
Berkeley, CA
Thank you, tshirt. But I've been to that page before posting the question, and the word "asymm*" does not appear. And according to the standard, the maximum bandwidth for upstream and downstream are only different by a factor of 1.4, a far cry from the 5x difference on my Business Class Deluxe, and even farther than the 10x of Business Class Deluxe 100.

Please review the requirements for a valid answer: You must be of DrDrew's caliber or better. A reference to Wikipedia doesn't get you there. Thanks so much, again.

ExoticFish

join:2008-08-31
Stuarts Draft, VA
And an attitude like that won't get you anywhere.

The technology behind our cable connection is what's causing the speed limitation is what I believe tshirt was trying to get across. The hardware itself COULD easily work on a symmetrical platform, cable and modems. As far as the head end hardware goes I'm pretty sure it could also, with changes of course.

Also when you look at it I think companies assume customers don't need as much upload bandwidth as they do download, which for the most part I'd say is true.
--
»www.VAJeeps.com
Expand your moderator at work


natedj
Elected
Premium
join:2001-06-06
Columbia, SC

1 edit
reply to bernardc

Re: [Speed] Download/upload ratio: Is it physics?

Actually companies know the importance of upload speeds that's why they limit it. With high download speeds you're a consumer that has the ability to download allot, with high upload speeds you can essentially host a server and actually make money with your bandwidth.
--
Good judgement comes with experience...Experience comes after bad judgements

bernardc

join:2009-12-13
Berkeley, CA
reply to ExoticFish
I'm looking for reliable, verifiable, solid information. If you don't have it, that's no problem. No one knows everything. I certainly don't, and that's why I'm asking. There's no need to reply if you don't have the answer the meets the needed specifications for rigor. I have no interest in offending anyone, really.


tshirt
Premium
join:2004-07-11
Snohomish, WA
kudos:5
Reviews:
·Comcast
reply to bernardc
said by bernardc:

You must be of DrDrew's caliber or better.

He won't talk to you, since you even SUGGESTED there MIGHT even be somone better.

Besides since you're in Berkeley, couldn't you just call Stanford.

bernardc

join:2009-12-13
Berkeley, CA
tshirt, I am so glad you have a sense of humor!!!

I really doubt anyone could be better than DrDrew.

Regarding Stanford, right! I heard they're the best and the brightest.


bobjohnson
Premium
join:2007-02-03
Orlando, FL
Reviews:
·AT&T U-Verse
·T-Mobile US

1 edit
reply to bernardc
I'm not DrDrew but i do know from a senior engineer why this is. There is no physical limitation of the hardware or cables but the specral inefficencies of Docsis up until 3.0 and the inability of cablecos to use certain parts of the upstream frequencies reliably. 6mhz for 30megs upstream is alot considering that is shared throughout your node (or multiple nodes) in smaller systems case.

Edit: I neglected to mention that once the analog channels are eliminated they will be able to use higher than 42mhz for the upstream but i quote "The system will always be lopsided for a number of reasons".

bernardc

join:2009-12-13
Berkeley, CA

Good info

Thank you, Bob. Now we're getting somewhere. I'll be interested to find out what specifically those reasons are for it always being lopsided.

You're right that the low frequencies upstream, being around 5% of the downstream, is an important limitation. The theoretical limit, called the Nyquist rate, is a bitrate of twice the frequency.


Caddyroger
Premium
join:2001-06-11
To the west
Reviews:
·Comcast
reply to bernardc

Re: [Speed] Download/upload ratio: Is it physics?

said by bernardc:

I'm looking for reliable, verifiable, solid information. If you don't have it, that's no problem. No one knows everything. I certainly don't, and that's why I'm asking. There's no need to reply if you don't have the answer the meets the needed specifications for rigor. I have no interest in offending anyone, really.

Then should list the thing that you looked. No one know that you all ready looked at Wikipedia.
--
Caddy


news

@videotron.ca
reply to bernardc
Before cable Internet, cable tv plant were one-way only, with RF amplifiers passing foward signals from 54 MHz to 550, 600 or 750 MHz depending of the cable plant. So they needed a 2-way plant for the Internet, they choose to pass the return data (from the customer to the cable headend) below 54 MHz, in the 5-42 MHz band, where there wasn't any tv channels . So the bandwith available isn't much, but anyway Internet was new, almost only downstream traffic... With time, they changed the upstream modulation from qpsk to 16qam and today some cable plant use 64 qam to be able to gain more speed and bandwith on upstream channels.

Today, catv manufacturers buid amplifiers with wider upstream band 5-55, 5-65 and 5-85 MHz, but most cable plant still only use 5-42 Mhz band.

»www.cedmagazine.com/articles/201 ··· m-speeds


telcodad
Premium
join:2011-09-16
Lincroft, NJ
kudos:17

4 edits
I'm not DrDrew either (and he and others more expert on this can correct/clarify what I say), and while "news @ videotron.ca" has touched on a lot of this already, here's my shot at this anyway:

The asymmetry between the much higher download speeds vs. the lower upload speeds is basically due to both the asymmetrical bandwidths and SNRs of the downstream (or "forward") path vs. the upstream (or "reverse") path.

Because of the original broadcast TV channel assignments in the US, the upstream path from the customer back to the cable headend has been limited to between 5 to 42 MHz (vs. bandwidths of 50 MHz or more that are currently being allocated to HSI service in the downstream).

This bandwidth-limited, low-frequency, upstream path is also susceptible to an number of impairments such as impulse noise, the "ingress" of interfering signals from outside the system (such as from CB and ham radio transmissions), and from noise and interference originating from inside subscribers' homes. This results in SNRs that are typically 6 dB or more worse than those obtained on the downstream path, so that only lower-order modulations that can tolerate the lower SNRs, like 16- or 64-QAM, can be used (vs. 256-QAM in the downstream).

Here are some recent threads in this forum that have discussed the limitations of the upstream path:
8 downstream 8 upstream modem? - »8 downstream 8 upstream modem?
4 Upstream Channels - »4 Upstream Channels

Here are also some other articles that discuss the issues with the upstream path and some things that can be done to reduce the asymmetry:
Why Is Cable's Upstream Path So Darned Ugly? - »www.translation-please.com/colum ··· umnid=97
Cable's Upstream Gap - »www.lightreading.com/blog.asp?bl ··· lr_cable
Will Anyone Switch On Cable's Upstream Booster - »www.lightreading.com/document.as ··· lr_cable

EDIT: BTW - Here are the other parts of Leslie Ellis's early articles about Cable's upstream path (and still fairly applicable for being over 10 years old!):
Peer-to-Peer Meets Cable's Upstream Path - »www.translation-please.com/colum ··· umnid=98
How Cable Gets More Stuff Upstream - »www.translation-please.com/colum ··· umnid=96


RR Conductor
NWP RR Inc.,serving NW CA
Premium
join:2002-04-02
Redwood Valley, CA
kudos:1
reply to bernardc
Good ole Berzerkley LOL I'm 130 miles northwest of you up here in Redwood Valley, on Comcast too, though the scenery up here is much nicer!

bernardc

join:2009-12-13
Berkeley, CA
reply to bernardc

More good info

Thank you news and telcodad for the very useful information and links.


bobjohnson
Premium
join:2007-02-03
Orlando, FL
Reviews:
·AT&T U-Verse
·T-Mobile US
reply to bernardc

Re: Good info

said by bernardc:

Thank you, Bob. Now we're getting somewhere. I'll be interested to find out what specifically those reasons are for it always being lopsided.

I'm just sharing the IM messages instead of passing it on in my own words.
Me (8/26):
so why would you guys not be able to get the upload symmetrical?
Phil (8/26):
It's not that we won't be able to, tech speaking. Docsis will get there if they want it to but the plant will not be able to. The system is shared with so many different things and rr is a small part. We already have symm fiber to business, whatever they want basically. Even when we go sdv and get rid of analog there will still be a lower usable amount of bandwidth for return bc of interference and other variables that are not in the labs but are in real life where they r deployed.
Phil (8/26):The system and specs were designed to send data to the sub and that's the way it will always be. We're basically retroing a 1 way design so eventually we won't be able to go any further. Sorry for writing a book. Have a good night
--


bernardc

join:2009-12-13
Berkeley, CA
Thank you again Bob for further good info.

(RR,
Do they still run the Fort Bragg steam train? Back in '89 it had the best whistle operator I'd ever heard.)


JigglyWiggly

join:2009-07-12
Pleasanton, CA
i thought it was just because there is less frequency devoted to the upstream?

harald

join:2010-10-22
Columbus, OH
kudos:2
reply to bernardc
An administrative decision in 1948 set the low end of the TV spectrum at 54 MHz. When 2-way cable systems were developed, that left only the frequencies below 54 MHz for the upstream and a guard band.

On the downstream side, the bandwidth is allocated between cable TV channels and Internet

On the upstream side the frequencies are allocated to cable modem upstreams (set-top box control data, FM radio, and cable telephone don't take up much bandwidth).

So the limitation is the amount of bandwidth available. The amount allocated for both up and downstream is decided by the cable companies.

Since the cable plant is strictly an analog system, changing the frequency break becomes very difficult.


DrDrew
That others may surf
Premium
join:2009-01-28
SoCal
kudos:20
reply to bernardc

Re: [Speed] Download/upload ratio: Is it physics?

said by bernardc:

Thank you, DrDrew for all of your fine work here.

You're welcome, but as you can see there are many others here who can answer the questions just as well.

A large part of the asymmetry of cable systems comes from the limited upstream bandwidth (50 vs 800 mhz) and the noise "funnel" effect of having hundreds of customers connected to a handful of upstream receivers.

The noise from those customers is additive and to fight the effects of that noise, throughput is traded for connection reliability in the form of lower modulation rates. Lower modulation rates have lower "bits per hertz", basically lower throughput per Mhz of frequency space, but they also have lower SNR requirements for clean data transfer. A single customer or cable fault can easily add enough noise to affect every other customer attached to the same upstream channels.

Downstream channels don't have the noise funnel effect as it's a handful of channels split-out to hundreds of customers. If the signal source is clean, amplifiers are running well, and the cable is in pretty good shape, end users will get a pretty clean signal so high modulation rates can be used. A amp or cable problem will tend to only affect that branch downstream of the problem. A single customer, being at the end of the line, will have a very difficult time affecting the downstream signal quality of others,
--
If it's important, back it up... twice. Even 99.999% availability isn't enough sometimes.

bernardc

join:2009-12-13
Berkeley, CA
Thank you DrDrew and Harald for the excellent information. Yes, there are some of your caliber, but they're not common. I was trying to preemptively filter most of the anticipated noise, and unintentionally ruffled a few feathers in the process.

Thanks again for this very informative discussion!


RR Conductor
NWP RR Inc.,serving NW CA
Premium
join:2002-04-02
Redwood Valley, CA
kudos:1
reply to bernardc

Re: Good info

said by bernardc:

Thank you again Bob for further good info.

(RR,
Do they still run the Fort Bragg steam train? Back in '89 it had the best whistle operator I'd ever heard.)

They sure do, and it's better than ever! I'll IM you all the goodies lol