It's really interesting to see the tech being deployed to fix this problem. But it also amazes me how complicated we've managed to make things. In the UK in particular - just a few years ago the vast majority of people received their TV through over-the-air broadcast. It's instant and scales infinitely, kind of beautiful simplicity, technically speaking.
Sports are a good test case, not just for latency but sheer demand. I worry that we'll face some kind of natural disaster in the not too distant future that'll make us wonder why we abandoned the resilience of OTA broadcast.
The interesting element of the online streaming is the rise of people watching on their phones/tablets, where getting a traditional OTA signal isn't possible.
Korea, too. Probably some other countries. I remember the PSP had an optional OTA antenna.
There have been several attempts at doing the same thing in the U.S., but none of them ever worked out for various reasons. Often it was because of the expense of upgrading the tens of thousands of TV transmitters across the country.
Yup, I remember when I thought having my Motorola RAZR was the epitome of cool, people in Seoul would flip up their seemingly dumber phones and stream crystal clear HD TV in the subway.
For South Korea it's not normal HDTV but a special broadcast for mobile devices, DMB [1]. It was only recently updated to support 720p in favor of 240p. They aren't picking up the domestic ATSC signals. The Japanese mobile video system is a similar separate broadcast.
Actors have vented their fury at two audience members who watched England's World Cup penalty shoot-out on their phones in the front row of a musical.
Titanic The Musical actor Niall Sheehy said the women "not only followed the penalty shootout on their phone, but also said 'yesss' on each goal scored".
He declared: "You are the most ignorant audience members I have ever had the misfortune to perform in front of."
You're blaming technology for social problems. For every person that interrupts a play, there are 10 people quietly watching on their train home... turning wasted time into time enjoyed.
Rude people are always going to be rude, and we shouldn't worry about the technology that enables them to do so. If they couldn't watch soccer during a musical, they'd just be coughing loudly or rummaging around in their pockets looking for a mint or having a too-loud "private" conversation with the person next to them.
I’m not confident it’s actually true that “phones have changed nothing”. They’ve certainly increased the variety of distractions one can access from a theater seat, as well as the ease of accessing them. But it’s also clear that most people have enough self-control to avoid using their phones for the duration of a musical, so it’s reasonable to blame those who don’t rather than blaming the technology. (This is distinct from the longer-term phenomenon of phone addiction, where there’s more room to blame the technology’s addictive qualities…)
Typical actors. A profession with a more entitled attitude is hard to find. You see it too in how often actors make grandiose political or social pronouncements and expect everyone to care.
They got paid to perform a play. There is no more obligation for the audience members to pay attention than there is for someone to eat all the food on their plate at a restaurant, or use a device they bought every day. If the women weren't disrupting the other audience members then the actors shouldn't have cared (and if they were, that's a problem for the house staff to sort, not them).
If actors understood that it's not the audience who is privileged to watch them act, it's them who are privileged to be paid to play make-believe on stage, the world would be a much better place.
Still do. While just over 1 million watched the last BBC england game on iplayer (before it crashed), at least 20 million were watching on BBC One.
Everything else makes perfect sense to broadcast, but live -- especially zero-latency live -- over unicast http calls? Ridiculous. A shame the BBC's multicast trials never took off.
In India , we had a 9.7 million concurrent streams record recently for a cricket game with a service called Hotstar. The stream at least for me was ahead of Cable TV, not sure if they delayed the TV broadcast to account for live streaming delay to make it appear live.
I love things that just work but it's a little unfair to compare radio technology that has been around for over 100 years and backed by huge public infrastructure investments with internet-based live streaming that has been around for (being generous) maybe 10-15 years at the most?
Not really when theres a good chance its ending up going through radio technology still anyways (wifi).
And its a matter of complexity - its obvious why OTA would have less problems to solve in order to get the same performance and behavior of something more complex.
OTA doesn't have less problems to solve. You have all the same problems of digital encoders etc (since all the media is digital anyway) PLUS the problems of transmission towers, solar events (I kid you not), wildlife, etc. You have the safety and cost problems supporting those hazardous and highly expensive equipment in remote locations.
IP streaming does actually solve a lot of problems. Plus these days OTA transmissions are all digital anyway. So serving that digital content via the internet is a natural progression. Sure things are superficially more complex at the moment but the technology hasn't matured yet.
I remember an anecdote from an old school broadcast engineer I knew about 15 years ago. He recalled some sales guy from Real coming in and saying "with this new box we can broadcast to 5 thousand people". Engineer pointed to a picture of crystal palace transmitter that happened to be in the room -- "with that we can transmit to 5 million".
I'm not convinced how muliticast works to mobile devices, but it's a painfully edge case. This type of event occurs in the UK once every 4 years (the olympics doesn't pull in these numbers -- except for 2012 when it was hosted in London), and even then only when England do well (so once every 20 years)
That said, ITV latency seemed far shorter than BBC latency, watching ITV player seemed acceptable. Apart from the lackluster commentary and the adverts.
* when the broadcaster screws up their RDS signal after a traffic report leaving you stranded on the wrong station.
* Or FM stations that run at different frequencies from different transmitters (this used to be a massive problem when driving long distances even just 20 years ago but now even the analogue car radios are digital devices that scan multiple frequencies to hop to the next frequency aired by a given station so you don't experience much cutout).
* Then there is the obvious issue of "dead zones" where the signals cannot reach (tunnels, some industrial locations, etc).
* Digital radio also suffers from frequent, albeit brief, cutouts while driving too but I've found that to have fewer dead zones than analogue.
* Bleeding used to be massive problem with analogue radio as well - this is another problem that was solved when radio started going digital.
So this is with 100 years of research and development into a protocol that requires an order of magnitude less bandwidth and yet most of the improvements have only been in the last 20 years when things like RDS (digital packets in analogue signals) and DAB have been introduced. Given that IP streaming video content isn't that far behind, just imagine how good it is going to be when that technology finally matures.
> but now even the analogue car radios are digital devices that scan multiple frequencies to hop to the next frequency aired by a given station so you don't experience much cutout
Wait, is this really a common thing? How would I know if a car radio does this, because I'm 99% sure my car doesn't.
Maybe it's not a thing any more? Or maybe it's not as clever as scanning multiple frequencies? I've not listened to analogue radio in the car for long distances in a while but I do remember there used to be a big problem with national radio spanning multiple different frequencies and your signal would drift out as you changed regions. This problem seemed to solve itself when RDS and "smart" radios started becoming a thing. However my experience is with video broadcasting rather than radio so I might have gotten some of the details wrong.
The name of the feature is usually AF (RDS), all my car radios since 90s have supported it and all multi-transmitter FM stations here make use of it. I'm in Finland - no idea if it is less common in US.
The problem is we only have a finite amount of spectrum available, and people like choice, so there are more channels than there are slots in the airwaves for them to be broadcast.
Just like the phones in the UK will, in the not too distant future, move from POTS to a fully packet switched network, so too will television, especially once at least superfast (24Mbps and up) broadband is available to all - there are already rumblings that Sky would like to move to IPTV instead of having to bounce their signals off a satellite.
As is so often the case in life, it's a case of squeezing a balloon. In order to get more of one thing, you have to have less of another.
In this case, to have more reliability in transmission, you have to have fewer choices.
I can't speak to the UK's digital situation, but in the U.S. there are 50 channels available in a theoretical city. Assuming that digital has fewer co-channel issues than analog (most of my TV experience is analog), that means 25 signals. On those 25 signals, some broadcasters squeeze up to 10 channels, which all look like garbage. For a decent picture, it's closer to four — maybe five.
So that means 100 to 125 theoretical OTA channels.
If people could actually receive 125 OTA channels with good programming on them, or even the same content as is offered on cable/satellite, I believe many would drop pay TV in a heartbeat, and wouldn't be forced to also scrounge for content on Netflix and YouTube and Prime Video.
If anyone has information on the theoretical maximum number of TV channels possible in a single location in the UK, I'd like to hear about it.
I’d be shocked if 125 good channels would convince people to give up on-demand streaming services. 125 really is not a whole lot when you need to satisfy everybody, and the need to obey someone else’s schedule really sucks.
Indeed, that's a bit silly. WE are calling these things 'broadcast', while most implementations are really unicast. This works really great for things like netflix, where you can start watching whatever you want, independently.
But in this case we really want actual broadcasts. Talk about square pegs and round holes...
VOD = video on demand, unicast is good. Live = stream, multicast is good. The 1.5 second multicast estimate for next Olympics from BBC sounds reasonable based on my experience working with signalling and now live straming systems.
Good read around the steps and impact around buffering introducing delay. My programmer brain said “sounds like the receiver should control flow into its encoder”... so is it that it’s too conservative a buffer, or the encoding algorighms actually need that much buffer to get economy?
That's not true in some ways, TV requires a huge number of transmitters and is limited to a few channels, where the internet has a virtually infinite number of channels.
Even now it's digital here in the UK I believe each of the 5(?) muxes can only support tens of channel each.
Airwaves TV might scale easily but can only send a proportionally small amount of information, but the internet send a vastly larger amount of data per second, but struggles when everyone wants the same data at once.
I don't disagree with your point per se but you are over simplifying things. Here is my take after working in the industry:
> It's instant
It isn't. There are still latency delays with digital encoders, transmitter hops, etc even with OTA signals. It's just the equipment has gotten so good over the years that the latency has been pushed right down but it did used to be a lot worse. The same will happen with IP streaming given time.
> scales infinitely
It doesn't. You need geostationary satellites / multiple transmission towers, and all that only provides you a finite number of channels. In fact digital streams are actually much cheaper to scale up, which is why so many broadcasters are moving into this territory (and why I currently have a job)
> kind of beautiful simplicity, technically speaking
Superficially yes. But if you spend any amount of time around OTA broadcasters (as I have) you'll see there is a hell of a lot of complexity hidden from sight. In fact since OTA broadcasts are still digital streams served from content stored in digital formats, a lot of the technology used for internet streaming is the same for OTA broadcasts. So there is an argument for the simplicity of just serving that digital stream via IP rather than radio waves (or are you suggesting we go back to analogue as well? Because that also introduces a lot of complexity but of a different nature, like having to giant robotic cabs filled with hundreds of magnetic taps that all have to be wound to the right point; and a dependency on the postal service delivering those tapes since you now cannot just download assets from the content owners).
I'm not saying there isn't a lot of new complexity with IP streaming but I do think it's more pronounced because:
1/ we work in IT so follow the technology more closely than we do for OTA broadcasting
2/ it's still a relatively young industry so there are a lot warts that haven't been cleaned up (much like how OTA broadcasting used to be a lot less polished when the technology there was still evolving).
Yeah, for events like the World Cup "scale" means sending the same broadcast to millions of receivers. That's already an unusual requirement, though. Major sporting events constitute the biggest use case, and perhaps even those will require something different in future if the next trend is something more interactive: VR or user controlled camera angles.
This is a legacy way of consuming material. More and more are using on demand first.
However the use case of live sport - especially football in the uk - is massively different. This means the platforms have to scale to 20 million plus live streams (that's say 100tbit and 5 second latency) for a once in 4 year event.
That's "just" a problem with unicast though. Lots of different media houses are working on getting multicast support over the internet, by which point IP streams will behave a lot more like OTA's one-to-many relationship.
I couldn't say how close they are to getting that working - nor if it even is possible - because I don't work on that specific side of things. But I do know there used to be multicast networks over the internet back in the 90s (IIRC David Bowie even streamed one of his concerts over it) so it strikes me as a solvable problem.
> It's instant and scales infinitely, kind of beautiful simplicity, technically speaking.
I'm not so sure. When I had analog over-the-air TV, in a non-capital city, I remember quite a few times when most of the channels would go down during a storm.
> just a few years ago the vast majority of people received their TV through over-the-air broadcast. It's instant and scales infinitely, kind of beautiful simplicity, technically speaking
"Scales infinitely" is underspecified. Over-the-air broadcast scales well with the number of people who want to receive a particular broadcast. It scales very poorly on some other dimensions:
- Over-the-air suffers from congestion when many parties want to send a broadcast simultaneously. There's only so much spectrum. In a wired network, you can just add more wires.
- Over-the-air doesn't scale well at all when people want to receive the same broadcast from farther away. In a wired network, you can use longer wires, or relay data from one wire to another wire. (Relays will work for wireless too, but they make congestion that much worse.)
I actually wonder why they don’t offer basic RTP streams. Sure, this new HLS thing is often better for mobiles and shit connections but when I’m on a 500Mbps connection in the office and want to watch the World Cup, couldn’t they just cut the HLS crap and give me raw RTP?
Wow, here in Brazil a 10-second lag is already unbearable... No matter where you are in a big city, you can hear the city erupt in different ways when the national team is playing
haha, that's true. it got to a point where my wife and i would close all the windows and watch the game using headphones on the tv because it was basically impossible not to hear "spoilers".
now one thing i'm impressed is that the internet transmission from SporTV had basically zero lag when compared to the TV one.
My cousin was nice enough to call at kickoff and paused his game until the FuboTV feed that I was watching caught up. Then we could text spoiler free for the rest of the game.
You get the same in the UK over terrestrial tv depending on whether you're watching on the old mpeg2 standard def channel or the newer mpeg4 hi def, the latter is around 8 seconds behind.
Don't get me on satellite latency 20 years ago where people used to make money on betting on horse races where the data feeds in betting shops were about 40 seconds delayed...
Let's not get ahead of ourselves here, BBC iplayer crashed during a ENGvSWE quarterfinal and YouTubeTV went down for 50 minutes during ENGvCRO just a few days ago.
>"If you talk about the next FIFA World Cup or next Winter Olympics I think we will have a 1.5 second latency," Mohite says. "But getting it to zero is going to take another two to three more years of time."
We know there is a roughly 5 to 6 seconds delay from Real life to TV Broadcast. ( Or may be from Betting company to TV Broadcast ) [1]. Which should have been enough time for OTT to catch up to OTA. Assuming if they could allow OTT to have 1.5 second head start of OTA.
And to be honest I am very much surprised how unprepared all these OTT players were ( If you live in a country that was well prepared and offer low latency streaming, please forgive me ). As if they just discover 30+ sec delay in OTT delivery is going to be a big problem in World Cup. And this isn't as complicated as a Low Latency Live Streaming to Worldwide audience which requires much more complicated Edge and POP measurement and design, they are basically streaming to audience within their broadcasting right which is all regional.
I have a hard time understanding why we cant do sub 8 seconds delay delivery today ( Compared to TV, not real life ). I think this is mostly due to manager and marketing department decide to throw in OTT as an added features for their customers to enjoy / upsell their package in the last minutes without fully considering, investing and testing their tech. We should have been aiming for 1.5 sec this World Cup. Not Next.
P.S - I do agree with another comment about OTA broadcast. I just wish our phone could receive those signal without much compromise. Instead it seems the world is moving towards IP based everything.
I watched the England semi final on a pub outdoor TV with approx 1.5 second discrepancy in latency between screens in different areas of the courtyard (no idea of the technical setup which achieved this). Learning the outcome of a shot/cross just as you watch the ball being struck is particularly weird (quite fun when a free kick is scored, arguably worse than a longer lag the rest of the time as there's little mystery about when/how something will happen)
It would be odd if the same pub had more than one satellite/cable connection, so that's presumably to do with different decoders inside each TV. But 1.5 seconds is a long time!
Whatever happened to multicast over the general internet? Seems like the perfect use case for it. Let the hardware do the copying instead of unicasting TCP connections or whatever it is they do.
AT&T did this for Uverse - it’s unicast for 5 seconds then it switches over to multicast. I assume there’s some key frames and h26x codec configuration that’s necessary.
Unfortunately multicast doesn’t work across the internet. It’s typically filtered or ignored I think.
BT in the UK do this. It's awesome, but doesn't fix encoding delays like iPlayer is suffering from. Plus, it's over their Network which is a subset of the UK public internet infrastructure.
I had an old blog page about how it was designed but I took it down. May be worth revisiting it. That said the general public prefer unicast so they can watch it when they want, which renders multicast a moot point.
Nagra sell a system that does it to ISPs around the world and have done for at least 10 years. I'd been hoping we'd see more of that in the UK, but there has been virtually zero movement probably largely because of our patchy broadband service. I'm assuming the Sky Q boxes are heading in that direction eventually.
Internet multicast turns out to be hard. Now that streaming video is popular enough maybe it will be deemed a problem worth solving. I'm sure it's solvable. But it's hard. Also the trend for most things is video on demand, simultaneous viewing of a live event is rare.
I've experienced the same problem 10 years ago watching the Olympics and I guess nobody solved it yet. Interesting thing is that 30 second delay is totally fine for 99.9% of the contents on TV except the games the whole country is watching.
Anecdotally, YouTube TV had ~15 second delay and Fox Sports Go had ~50 second delay compared to what I see on the cable TV.
I use the Forza ios app, and during the last few games I had to literally hide my phone to avoid its notifications. The lag was more than a minute on England matches.
To be fair, apart from the lag and that crash at the end of one game, the BBC stream was a rock-solid HD, and it almost never buffered. ITV instead was low-res crap, stuttered all the time, and even had compulsory ads. Both streams got worse as the tournament neared the end, of course; in the first rounds the latency was only a few dozen seconds.
The easy solution would be to delay the TV signal by as much. This is only a problem during the world cup, and only because your neighbors will yell goal before you can see it. Delay artificially the TV and the problem is gone.
Most of the services were around 30 second delays, with the best being around 15 seconds and the worst around 60 seconds.
At this point there needs to be a technology shift if its going to get any better than about 10 seconds.
It even varies by device. iOS defaults to 10 second chunks, but you can reliably get it down to 6 seconds. You generally need to store at least 2 chunks or you'll have bad buffering.
I haven't been watching the football, but I have been streaming another live event for the past two weeks: Wimbledon. My advantage comes from two things: I'm blind, and the BBC's CDNs have audio-only variants for every HLS stream. So I've been able to cut out the huge amount of bandwidth that would otherwise be consumed by streaming HD video, run the audio-only variants through FFmpeg and seek right to the end until the stream buffers. I'm usually ahead of the official Wimbledon live scoreboard, and the same coverage on cable TV, by a few seconds.
Yup, this sums it up. Now that we're in the second week of the tournament there are less matches to cover and the radio becomes a good option. Before that, there are too many good matches on outside courts to stick to the main action.
Mildly OT, as you’re a blind tennis fan, have you ever tried the adapted version of the sport for blind and partially sighted participants? If not, and you’re interested, I might be able to point you in the right direction depending on where you’re based.
Sure, I played for a short while in Leeds. Good exercise but not enough opportunities to practice, leading to nobody really developing their skill unless they had the time and money to travel around the country to different clubs. Realistically there's not a lot to be done about that but I found it frustrating. Happy to hear if anything has changed since.
I used illegal streams this world cup. They seemed to have sub millisecond latency i.e got the goals in near real time. Maybe, BBC & iTV should ask what the pirates are using.
They’re most likely grabbing the feed directly from a TV tuner and then sending that over RTP. While it won’t be sub-millisecond, it’s definitely a lot faster than HLS.
With the amount of content that internet have, torrent of movies with incredible quality, beautiful documentaries, Netflix, youtube with your favorite tv shows, TV turned obsolete except from sports that are transmited live. So i feel like the main thing that tv has to got right its reduce to the maximum the delay time over quality of image.
It's not a problem that can be fixed. The endpoints - encoding at the transmitter and decoding at the device - can be with the efforts underway made somewhat equivalent between both modes. However, the problem is the path between them - the Internet was never built for realtime delivery of media streams. The many layers and devices along the route add delay, and on top of that network congestion can add more. Unless the path is more or less dedicated (as happens in broadcast, satellite or cable TV), there is no way to ensure realtime delivery.
"RTP does not ensure real-time delivery. So how come it is called a real-time protocol?
No end-to-end protocol, including RTP, can ensure in-time delivery. This always requires the support of lower layers that actually have control over resources in switches and routers. RTP provides functionality suited for carrying real-time content, e.g., a timestamp and control mechanisms for synchronizing different streams with timing properties."
]
y, it's funny when watching NFL over the air if I tweet my friend about a TD he usually finds out from me since over the air is a few seconds ahead of what he's seeing on DirectTV.
Sports are a good test case, not just for latency but sheer demand. I worry that we'll face some kind of natural disaster in the not too distant future that'll make us wonder why we abandoned the resilience of OTA broadcast.
reply