Hacker News Reader

Comment by jcrawfordor | Hacker News Reader

Most cable carriers, prominently Comcast, have issues with legacy channel allocations (e.g. for the On Demand feature for legacy Motorola STBs) that conflict with the channel allocations required for the higher upstream speeds supported by DOCSIS 3.0 and 3.1. This results in a de facto situation of upstream speeds being constrained to only a very small portion of the up to 2gbps upstream capability of 3.1, due to only having typically 3 QAM channels for upstream.

Unfortunately, eliminating these issues requires large-scale replacement of not just equipment in the field (nodes and amplifiers which are broadly all being replaced anyway to support Node+Zero architecture) but equipment in customer homes: the legacy STBs. Cable carriers have found in the past that getting people to turn these in to swap out for newer models is excessively expensive in customer goodwill, and less so cost, so there's a major technical disincentive to such a change.

View on HN

>legacy channel allocations

if this issue came from a cable provider it sounds like malarkey.

Vector Orthogonal frequency-division multiplexing can be used with traditional OFDM receivers even if the channel arrives as much as 180 degrees out of phase for a traditional allocation. it permits vector subchannelling for modern receivers to get to "new" broadband regulated speeds. when the vector size M=1, VOFDM returns to OFDM and in other conditions it just turns back into SC-FDE Single-carrier (frequency-domain-equalization.)

https://en.wikipedia.org/wiki/Orthogonal_frequency-division_...

>Cable carriers have found in the past that getting people to turn these in to swap out for newer models is excessively expensive in customer goodwill

cable companies are hardly a station of the cross. AT&T and Comcast are renown for forcing customers to pay for modem upgrades to support digital boxes. XFinity X1 boxes are basically outlining how cable TV intends to remain set-top relevant in the 21st century by transitioning the box to 802.11 and ethernet. its also hardly difficult to see a scenario where not only your cable provider forces you to change, but the FCC strong-arms you as well. in 2008 the FCC basically "rechannelled" the VHF and UHF fast-scan TV spectrums to make way for digital TV.

IMO nows the time if they want to do it. cable TV is basicaly a dead animal and needs something, anything to keep it alive. full digital gigabit transition FTW.

The quality of the channel often does not permit the use of OFDM until distribution amplifiers are replaced. Unfortunately this is a totally different issue for the upstream and downstream directions as well, which is why several carriers, including Comcast in many markets, currently do OFDM downstream and only QAM upstream. The downstream issue has, on the whole, always been easier to fix due to the design of the outside plant, which is why 1gbps down has become a standard cable offering in DOCSIS 3.1 markets. In the case of upstream channels they are constrained by a variety of different equipment, some of which predates the use of QAM for all metadata.

Comcast is beginning a (presumably long) process of transitioning away from conventional cable television and to completely OTTd television service, which is essentially IPTV over DOCSIS, branded as Flex. This seems to be their long-term strategy and will allow a radically higher degree of flexibility in how they run their infrastructure. I would expect this to become the norm over the next ten years or so as DOCSIS is increasingly viewed as the primary purpose of the cable infrastructure.

Fascinating! I had no idea there was a delineation between upstream and downstream encoding albeit its a fairly obvious condition in retrospect thinking of the upload caps.

It’s a fairly obvious distinction considering that you’re dealing with a system that was initially designed for one way communication. The cable system is not an Ethernet network with multiple pairs of conductors for each direction, all hooked up to a switch. It’s effectively a single shared bus, which back in the day was shared between hundreds of houses. Remember, it was initially designed to just get OTA signals into TVs over cable.

It's also important to remember that fiber internet, in the sense people are usually talking about, is fundamentally the same. PON (passive optical networks) are passive in that a single fiber is shared between multiple houses using passive splitters. This means that the fiber is a shared medium and upstream and downstream, and multiple users, must all be managed based on TDM. In the widely used GPON, Gigabit PON, that shared medium operates at a gross rate of 1gbps as the name implies. The exact same issues of coordinating the shared medium between users exists for fiber last-mile technologies but is generally better managed because fiber carriers don't have nearly as much legacy to contend with and so universally use more modern scheduling and allocation methods.

From a practical perspective, any method of last-mile internet delivery will probably rely on a shared medium. The cost of "home run" cable routing is just too high, the telephone network which was originally structured this way has invested a great deal to move away from it wherever possible. There is no widely-used last mile internet technology today which does not have bandwidth contention between users. Fortunately, in practice, a 1gbps gross shared medium is usually sufficient to offer "gigabit symmetric" service to a neighborhood of subscribers without complaints, although there is some debate about whether or not popular 4K video streaming will change this.

Sounds like an operational risk on an RCM somewhere in the ivory tower. Instead it's cheaper to stay competitive through regulatory capture. I wish bribing politicians was more expensive than providing a better product </r>

So we need to elect greedier politicians who wont accept a pittance as a bribe. </s>

Or just create a system that selects from a pool of already independently wealthy candidates.

People don't become "independently wealthy" by saying no to money, they do it by grasping at every bit they can. Wealth is positively, not inversely, correlated to greed. We literally just saw what happens when you put an "independently wealthy" in the oval office, he took a 10 million dollar bribe to greenlight an oil pipeline.

Cap political donations per person and ban corporate donations. Money is not, and should not be legally considered, speech.

So... we should select politicians from the poor, then?

> Cable carriers have found in the past that getting people to turn these in to swap out for newer models is excessively expensive in customer goodwill, and less so cost, so there's a major technical disincentive to such a change.

I wish carriers would have a different brand for "enthusiasts". The type of customer who won't mind going to his back-panel and swapping equipment if the rest of the network supports it.

They sort of support "pro" users, but at a premium. Comcast's 1GB package is somewhere between $80 and $100 depending on contract duration and payment options. 2GB internet is $299/mo plus $1k in installation fees! But at least in the review I read it's symmetrical.

https://www.xfinity.com/support/articles/requirements-to-run...

https://medium.com/@Gtwy/comcasts-2000mbit-fiber-to-the-home...

I have it and it’s glorious. Spendy, but you’ve got your own 10g port on a switch at Comcast’s headend. And it’s actually 3 gbps total (between the 1g and 10g ports on the switch).

It sounds so much better than dealing with Comcast's cable plant. I keep a set of filters next to my modem so I can fiddle with the signal strength as large temperature changes affect it which in turn causes wild swings in my upload and download speeds. A truck roll might fix it once and for all but in the age of Covid that can't happen. Having fiber to the premises sounds heavenly next to that.

i believe the 2gb symmetrical plan is actual fiber to the home, rather than the cable of the 1gb plan. but im not an expert by any means

According to the second article I posted where a medium.com user shared his experience, yes Comcast brought fiber to his premises.

For once it's nice to live in old Europe :)

I have ftth from Fastweb in Italy and I get 1 Gbps down/200 mobs up for 24,95 €/month (with unlimited phone service included).

Apparently they're also going to bump download speed to 2.5 Gbps. You still have only 1gpbs ethernet ports on the provided modem/router, but the core idea is to give at least 500mbps to each client.

The broadband situation varies greatly across the US because we're such a big, sprawling country. Much of my city has access to symmetric 1G/1G connections, for example. Google has even been offering free fiber internet to public housing and select affordable housing buildings ( https://support.google.com/fiber/answer/6349491?hl=en ). Meanwhile, my friend who doesn't live very far from me is limited to DSL because he lives in a sparse location where houses are few and far between, making it prohibitively expensive to run new broadband infrastructure to everyone.

Mandating high speed internet access for everyone sounds great until you look at the costs for running new infrastructure to some of our more remote, rural cities.

>Mandating high speed internet access for everyone sounds great until you look at the costs for running new infrastructure to some of our more remote, rural cities.

The same could have been said for water, gas, sewage, electric, and mail.

>Mandating high speed internet access for everyone sounds great until you look at the costs for running new infrastructure to some of our more remote, rural cities.

Yes but broadband companies have already been given tax breaks and subsides to do that very work. They just haven't because (as an example) AT & T argues DSL still qualifies as sufficient for internet access. The original article was about upping the speed required to meet the definition of broadband from 25 down 3 up to 100 symmetrical. I'm not sure that'll actually make a difference unless broadband providers are required to install that new infrastructure.

When you’re talking about subsidies, cable companies are in a totally different boat than phone companies, because they’re regulated under an entirely different section of the communications act. It wasn’t until the Obama administration that broadband even came within the ambit of federal subsidies. Until then, the money was just for phone.

That's why Starlink will make sense mainly in the US, as there's only an antenna for the user to plug, there's no landline to extend. I estimate AT LEAST 25 million potential American customers that are in the same situation as your friend

One thing that really can bog down these infra things in the US (ya know, besides the general "got mine. screw you!" attitude of the oligarchs who rule this place) is the urban sprawl. On the same plot of land, you can have 8-16 housing units in a low-rise apartment building paying that 24.95 a month, whereas in the US that would be a single-occupancy detached home most likely.

Can you get good fiber (or internet in general) outside major urban centers in Italy?

> I wish carriers would have a different brand for "enthusiasts". The type of customer who won't mind going to his back-panel and swapping equipment if the rest of the network supports it.

Problem with cable TV: it's a shared medium. Which means that for upgrading one pro customer, they'll have to either lay a new cable or get all of the normies on his cable branch to new STBs.

> I wish carriers would have a different brand for "enthusiasts"

Enthusiast customers aren't a good money maker for large-scale companies. The challenge is that they demand more performance, but aren't willing to pay much more for it. Also, for every 1 homeowner who can actually handle equipment swaps without an installer, there are probably 2 to 5 more who think they can do it but will blame the company, leave bad reviews, and complain online the second that they get stuck.

As a result, companies tend to stick to a generic consumer option and a separate, higher-priced business option. If enthusiasts were really willing to pay higher prices, they'd just grab the business package and accept the higher price.

Channel allocation suggests they can choose this. So stop offering fake 1Gbps and allocate some of that downstream to upstream instead?

Comcast not only has the man power but money to make this happen.

and decades of lead time since they've been doing cable internet

Supposedly, DOCSIS 4.0 FDX is the solution, and hopefully it's coming soon.

https://twitter.com/ArtemR/status/1267220689548791809

isn't this what that $30 a month rental fee is for?

Reply on news.ycombinator.com