This seems like a puff piece for cable to continue to justify why they continue to offer abysmal uplink speeds instead of improving infrastructure.
It’s unsurprising that the uplink caps at 5mpbs, it’s a 50/5 connection; but the article just wants the reader to “take their word” that the video stream look good without also comparing other connections (the apps themselves might self regulate) or latency (the apps might employ more buffering on a slow connection)
This is literally a lobby group composed of legacy coaxial copper plant last mile operators with a vested interest in squeezing every last ROI dollar out of docsis3, so your statement is quite correct.
The South Park nipple rubbing cable tv operators but in real life.
Eh, I wouldn't call CableLabs a "lobby group" although this is 100% a puff piece justifying the continued use of DOCSIS. They do real R&D and certification, and are the ones who actually develop DOCSIS and other standards for cable.
DOCSIS 3 maxes out at 1000/200, so if they would just provide that instead of trying to claim a 5Mbps upload is "broadband".
Cable is a perfectly viable technology, with DOCSIS 3.1 and 4, for at least another decade, probably more. Their own lack of infrastructure investment is coming to bite them in the ass, fast.
For clarification: I work for an ISP/telco that manages legacy copper POTS/xDSL, cable, fiber and cellular networks. While we are actively expanding fiber from the backbone out and doing a lot of FTTH by now, cable is still a solid workhorse, but it takes an operator who actually cares about maintaining the infrastructure.
If you try to squeeze in as many endpoints as possible, sure.
With a competently designed network, you wouldn't have anywhere near that many in a segment.
I'm on a 300/60 cable connection via my employer, and I have never had my connection test below 360/70. This is in an an area densely populated with apartment buildings, where cable is still the best option for most people, with some fiber rolling out.
I am however moving to a 100/100 fiber connection (plenty fast for my needs, with the faster upload speed), due to a change in employment that is not of my own choice. So I don't consider myself biased in favor of my current employer ;-)
We support up to 1Gbps/100Mbps as a standard subscription on cable, with symmetrical connections available for an additional fee, except on 1Gbit, where the fastest upload available is 500Mbps.
Cable is alive and well, if properly maintained and upgraded.
I'm guessing it's a relatively small network, considering what I know about DOCSIS and how much would have to be upgraded to support uploads like that.
It covers the majority of urban and suburban Denmark, plus a majority of exurbs, anywhere an "antenna association" used to or still exists today and almost every apartment building. They are connected with coax through either us or one of our competitors, and there's mandated open access to coax.
That would most likely be because of low latency and jitter. The app is in essentially the environment it was developed in and does not have to put effort into correcting errors that itself cause pauses in the visible video, since it tries to hide the parts where artefacts appear.
Ten years back I was told America likes to zap quickly through tv-channels with all the artefacts involved while Asia preferred a slower and artefact-free experience. Analog and keyframeless digital codecs are faster but not so bandwidth efficient.
Without providing details of the cameras in the test laptops these numbers don't have a lot of meaning. The typical laptop has a potato quality webcam that strains to push 30fps at 720p, many can only manage 480p video even if they support higher resolution stills. There's a small number of laptops with higher quality cameras (Windows Hello if usually a checkbox feature of these) that will do 1080p video.
So saying a "mix of laptops" and "nothing special" isn't actually that helpful. Having ten laptops uploading their potato quality video over wired Ethernet without issues is a completely unsurprising result.
A 5mbps upstream connection is much less useful for those that have better quality cameras available. Most phones and tablets have much better front cameras than laptops and desktops. Even the iPad mini can do 1080p video on its front camera. A year into the pandemic and good webcams are still sold out so there's a non-zero population with 1080p webcams they're using for conferencing. A 5mbps upstream is a joke for people doing Twitch et al live streams.
Sorry cable internet providers, your terrible service for obnoxious prices is not a good value. A 5mbps upstream connection is absurdly low in general, let alone in a world where people are spending hours a day video conferencing.
I'm in no way defending US internet providers, but is 5 Mbps really that low for video-conferencing?
Up until late last year, my parents' connection was an ADSL that could only do 4-5 Mbps download. I could watch 1080p videos on Amazon Prime without any issue, and the quality was much, much above what I can possibly expect a good webcam to provide. Hell, I remember watching Star Trek in 480p on a MacBook Pro Retina in bed and being impressed by how good the image quality was.
I understand that real-time compression isn't as good, that movies have better cameras, lighting, etc.
But that's the thing, though. Maybe instead of pushing a ton of random garbage pixels per frame, a lower resolution will be just as good.
Instead of pushing for a 4K 60 fps or more webcam, let's make a damn good 720p one. Maybe add some diffuse light to it, like the Razer Kiyo. It will probably improve the perceived quality much, much more than an ultra high def one fighting with backlight and other suboptimal environment issues.
I'm not saying video conferencing needs 4K 60fps video. I'm saying that 5Mbps might suffice for the shit quality video most laptops will generate, it's insufficient for the much better quality video even modest 1080p cameras will generate.
Most laptops have terrible cameras. Their best output can be encoded at a few hundred Kbps and would be neigh indistinguishable from video captured by the same camera directly to disk. A good 720p or 1080p camera needs quite a bit more bandwidth to maintain good visual quality. If upstream bandwidth is constrained the software will dial back the visual quality to potato levels. For anyone that needs non-potato quality this is problematic.
Cable Internet companies patting themselves on the back for handling a dozen potato streams is just sad. Not everyone is ok with such quality. In an online classroom setting, the students don't necessarily need great quality upstream video but the instructor trying to capture a white board or switch between live video and a detailed presentation does want that high quality upstream video.
Video quality from streaming services is immaterial here. There's a lot of quality/bitrate tweaks that can be done with offline encoding that are not workable for real-time encoding.
I think that’s part of the OPs point–this article tells us almost nothing usable other than its “good enough” and when that type of statement accompanies no information about actual quality, it’s useless and is almost certainly a puff piece funded by isps.
My dad considers the computer he bought in 2005 “good enough” but most people who are tech literate and who live on technology would consider that to be painfully ridiculous.
I personally don’t understand the claimed need for such high bandwidth for video calling and general business use.
My work laptop has forced software updates, so at the start of lockdown, to protect my home internet connection I hard capped it to 4Mbps in both directions at the switch. I’ve been having ~back-to-back Meet video calls all day for an entire year without exceeding this or having any problems.
I think the quality of the connection (consistency, upstream contention, jitter, packet loss) is probably far more important for most practical business applications (ie excluding HD Netflix, game downloads etc).
There's a concept of "contention" when it comes to consumer broadband.
Your 50/5 broadband isn't dedicated to you, instead you share a single up link connection.
Depending on where you live and how they've extended it, you could have terrible ratios like 1:20.
This is often where you get the issues with connection consistency.
However consumers have no way of seeing what their ratio is.
So I'd argue getting higher bandwidth connections to households means that this would give a reason for ISPs to upgrade their infrastructure.
Also as consumers we should demand more not less from service providers.
50/5 is terrible. 625 kilobytes a second upload, that's objectively horrible internet.
If you're sharing a home with other people, you need to say to them "I'm going to have a conference call, could you not use the internet"
That is ridiculous.
Video conference calls can have higher quality streams as well.
But also there are plenty of industries where they aren't just pushing text to the internet.
What about creative fields where they want to upload a multi megapixel image they're going to get printed on a billboard?
Or a game developer who wants to iterate on assets?
High quality remote desktop streaming to a high end PC in a data centre?
There's a lot of business applications which aren't just email and code which would benefit from a good internet connection.
Where one person will have to wait seconds for an upload, and someone else has to wait minutes or hours.
Like I have a 1000/1000 connection in the UK, that has a low contention where I get great consistency of connection. And it costs me the equivalent of $70 a month.
It enables so much in an internet connected world. And I think as consumers we should be demanding that.
Not arguing that a connection that'd be impressive in 2005, is acceptable.
As the article shows, you can run 10 concurrent video calling sessions over 5 Mbps upload. That’s hardly “I’m going to have a conference call, could you not use the internet”.
There’s plenty of use cases for higher upload bandwidth, some of which you mentioned, but VCs on current generation laptops isn’t one of them.
A bigger issue is over reliance on WiFi in dense neighborhoods or through multiple walls. A better or more widely deployed in-unit wired network technology would be more impactful for a lot of people. I wish cat6 Ethernet was standard in construction, but in the US at least it seems to not be still. Our building was renovated in 2009 and has plenty of coaxial connections: 3 in the living room and 1 per bedroom, but no Ethernet. I’ve been slowly adding Ethernet to the coax plates, but it’s a real pain in the tail to do in an already built building.
Is there any examples of the quality of the video? If someone's face is a handful of pixels and them presenting their screen is a blur or super laggy, is that good enough?
What happens if one person makes a Google Duo call? That's in HD.
Or if one person turns on their console and starts uploading a save file to the cloud?
You can run networking over your existing coax wiring, with MoCA 2.5 you can get over 1Gbps, which does require you to buy adapters, but you don't have to pull new cabling.
This only works because video conferencing generally uses low resolution (320p/720p), low framerates, and terrible bitrates (that adapt downward as your connection lags).
It's really annoying that with all participants on symmetric gigabit connections we can't have 1080@30fps or 2160@30fps on any major video platform.
This is definitely a "5mbps ought to be enough for anyone!" puff piece (as another commenter pointed out).
I wonder why that is (no 1080@30fps or 2160@30fps) - is it back end bandwidth? For Teams at least I guess that everything is aggregated in the MS Datacentre..
Most people don't care about resolution. Try iMessaging or Signal messaging some 4k video from a $1400 iPhone sometime, it gets silently and automatically recompressed to potato before upload, and the hundreds of millions of users don't seem to notice or care.
Doesn't make sense for those companies to pay money to increase resolution and framerate when most of their customers are on shitty wifi with shitty cameras and wouldn't notice a quality increase anyway.
Had they run 5/5 Mbps or even 9000/5 Mbps, then they would of hit the same ceiling of active VC connections.
Biggest issue with video over networking would be bit errors as those micro delays of retransmission of if just sending without error checking, you will notice the impact. Audio a single bit error and you won't notice, networking you won't notice as all transparent or insignificant to notice. But with video, artefacts stand out hard and a single bit error can do that.
I did Video-conferencing in the late 90's and 128Kbps H320 VC link's worked for most and if you wanted to roll-out the red carpet you would bond 6 channels for 385kbps and that was with dedicated ASICS using codecs far weaker than what you can run in the background upon today's disposable CPU's and that shows how far things have moved in raw processing power. Back then to do conference calls you would need a MCU (Multipoint Control Unit) and those would be many hundred of thousands. Today, you have all the raw CPU power to do that on your mobile phone.
All tests done over wired Ethernet on a controlled cable system node. I’m not at all surprised that 5 Mbps was sufficient.
Most people’s issues with poor video call quality aren’t related to the last mile bandwidth, but instead to poor WiFi connections in the house. I’ve been pulling Ethernet cables through my house to fix this and it’s made a big difference. It also makes everyone else’s WiFi connections marginally better when I’m no longer competing for the airwaves.
A secondary issue is overloaded neighborhood cable termination nodes. Comcast is pretty slow at node splitting, leaving nodes overloaded for months on end with no alternatives available for people in areas where it’s Comcast or bust.
There’s issues aplenty, but it’s not with sufficiently provisioned 3 or 5 Mbps upload bandwidth.
This feels like convenient timing with the all of the recent talk of raising broadband standards. Its also convenient that this seems to not actually measure quality by anything other than “good” which sounds an awful lot like the language used repeatedly by shady isps who were fighting against this last time. I clearly remember their constant usage of “...good enough...”
They give us nothing, not even simple bitrate or resolution and fail to tell us really anything.
I’ve grown incredibly skeptical of anyone who says internet speed of X is “good enough.”
It wouldn’t shock me in the slightest if we were to learn this study was funded by AT&T or Verizon.
I moved away from cable because my area had poor capacity planning (and it's generally a crappy technology) - if someone in the house was on a video call (edit to say: any upload really) - ping times to any external host (like 1.1.1.1) would go from 20 ms (the minimum obtainable) to 60-100ms with jitter
The by far biggest factor on the quality of video conferencin is the microphone, speakers and camera.
If you participate in video conferences, do yourself a favor and buy a "HDMI capture USB stick" so you can hook up your "real" camera to your PC and get a great image.
They cost less than $20, work with Windows, macOS and Linux and don't even require any drivers.
Some of the fairly recent cameras also support a USB connection directly.
By "camera", do you mean a photo camera? If so,
won't such a setup put the sensor under considerable stress if you do video conferences frequently? I'm not sure, it seems to me that any "1080p webcam" would achieve similar results at a similar (maybe slightly higher) price, with less risks...
“1080p webcams” still have pretty poor focus, color, and sharpness. Any dedicated digital camera, even a lot of higher end smartphone cameras, will do better than any webcam.
Considering what my 35mbit upload looks like with screensharing I guess they like seeing pixels flip every few seconds and seeing people's faces not move in conference calls. No word about quality or what is considered acceptable and I guess it was avoided because it will be abysmal. There is a big difference between "it works" and "it's workable".
A 720p upstream only needs like 0.5 Mbps upload. If you can’t get good quality on your 35 Mbps upstream connection, then something else is wrong with your network.
Could be your local network, could be a poor cable somewhere between you and the CMTS, could be the node is overloaded and doesn’t have enough trunk bandwidth, could be poor peering. Could be lots of things, but it’s not your provisioned connection.
Provisioning a higher bandwidth connection on the same infrastructure wouldn’t fix any of that.
I had significant upstream issues that were caused by frayed coax cable coming from the street pole to the house. Technician replaced the cable and it’s been flawless ever since. Google Meet calls work great over my 75/15 connection and wired Ethernet.
They measured video quality subjectively. I don't think you can just take the researchers at their word that it "looked good enough". Any rigorous video quality test would use at least some kind of objective numeric measurement.
Which is generally misleading as human attention focus's on a golfball size area of the screen at any moment, and codecs take advantage of this. So you might get a higher numeric score for a much poorer subjective experience.
Yeah and the handovers are awkward. I haven't personally tried to conduct a conversation, but imagine that would be worse because the interchanges are more frequent, less predictable, and an important cue. But at 600ms, the cue is too late, so it won't work. I know that as theory, and wonder what the actual experience is like. Can people adjust, or is it unusable?
Would it be asking too much to expect an employee to keep in touch with work via videoconferencing regularly through geostationary satellite? Or is it bearable?
It’s unsurprising that the uplink caps at 5mpbs, it’s a 50/5 connection; but the article just wants the reader to “take their word” that the video stream look good without also comparing other connections (the apps themselves might self regulate) or latency (the apps might employ more buffering on a slow connection)