Video: As Time Goes by…Precision Time Protocol in the Emerging Broadcast Networks

How much timing do you need? PTP can get you timing in the nanoseconds, but is that needed, how can you transport it and how does it work? These questions and more are under the microscope in this video from RTS Thames Valley.

SMPTE Standards Vice President, Bruce Devlin introduces the two main speakers by reminding us why we need timing and how we dealt with it in the past. Looking back to the genesis of television, points out Bruce, everything was analogue and it was almost impossible to delay a signal at all. An 8cm, tightly wound coil of copper would give you only 450 nanoseconds of delay alternatively quartz crystals could be used to create delays. In the analogue world, these delays were used to time signals and since little could be delayed, only small adjustments were necessary. Bruce’s point is that we’ve swapped around now. Delays are everywhere because IP signals need to be buffered at every interface. It’s easy to find buffers that you didn’t know about and even small ones really add up. Whereas analogue TV got us from camera to TV within microseconds, it’s now a struggle to get below two seconds.

Hand in hand with this change is the change from metadata and control data being embedded in the video signal – and hence synchronised with the video signal – to all data being sent separately. This is where PTP, Precision Time Protocol, comes in. An IP-based timing mechanism which can keep time despite the buffers and allow signals to be synchronised.

Next to speak is Richard Hoptroff whose company works with broadcasters and financial services to provide accurate time derived from 4 satellite services (GPS, GLONASS etc) and the Swedish time authority RiSE. They have been working on the problem of delivering time to people who can’t put up antennas either because they are operating in an AWS datacentre or broadcasting from an underground car park. Delivering time by a wired network, Richard points out, is much more practical as it’s not susceptible to jamming and spoofing, unlike GPS.

Richard outlines SMPTE’s ST 2059-2 standard which says that a local system should maintain accuracy to within 1 microsecond. the JT-NM TR1001-1 specification calls for a maximum of 100ms between facilities, however Richard points out that, in practice, 1ms or even 10 microseconds is highly desired. And in tests, he shows that with layer 2, PTP unicast looping around western Europe was able to adhere to 1 microsecond, layer 3 within 10 microseconds. Over the internet, with a VPN Richard says he’s seen around 40 microseconds which would then feed into a boundary clock at the receiving site.

Summing up Richard points out that delivering PTP over a wired network can deliver great timing without needing timing hardware on an OPEX budget. On top of that, you can use it to add resilience to any existing GPS timing.

Gerard Philips from Arista speaks next to explain some of the basics about how PTP works. If you are interested in digging deeper, please check out this talk on PTP from Arista’s Robert Welch.

Already in use by many industries including finance, power and telcoms, PTP is base on IEEE-1588 allowing synchronisation down to 10s of nanoseconds. Just sending out a timestamp to the network would be a problem because jitter is inherent in networks; it’s part and parcel of how switches work. Dealing with the timing variations as smaller packets wait for larger packets to get out of the way is part of the job of PTP.

To do this, the main clock – called the grandmaster – sends out the time to everyone 8 times a second. This means that all the devices on the network, known as endpoints, will know what time it was when the message was sent. They still won’t know the actual time because they don’t know how long the message took to get to them. To determine this, each endpoint has to send a message back to the grandmaster. This is called a delay request. All that happens here is that the grandmaster replies with the time it received the message.

PTP Primary-Secondary Message Exchange.
Source: Meinberg [link]

This gives us 4 points in time. The first (t1) is when the grandmaster sent out the first message. The second (t2) is when the device received it. t3 is when the endpoint sent out its delay request and t4 is the time when the master clock received that request. The difference between t2 and t1 indicates how long the original message took to get there. Similarly t4-t3 gives that information in the other direction. These can be combined to derive the time. For more info either check out Arista’s talk on the topic or this talk from RAVENNA and Meinberg from which the figure above comes.

Gerard briefly gives an overview of Boundary Clock which act as secondary time sources taking pressure off the main grandmaster(s) so they don’t have to deal with thousands of delay requests, but they also solve a problem with jitter of signals being passed through switches as it’s usually the switch itself which is the boundary clock. Alternatively, Transparent Clock switches simply pass on the PTP messages but they update the timestamps to take account of how long the message took to travel through the switch. Gerard recommends only using one type in a single system.

Referring back to Bruce’s opening, Gerard highlights the need to monitor the PTP system. Black and burst timing didn’t need monitoring. As long as the main clock was happy, the DA’s downstream just did their job and on occasion needed replacing. PTP is a system with bidirectional communication and it changes depending on network conditions. Gerard makes a plea to build a monitoring system as part of your solution to provide visibility into how it’s working because as soon as there’s a problem with PTP, there could quickly be major problems. Network switches themselves can provide a lot of telemetry on this showing you delay values and allowing you to see grandmaster changes.

Gerard’s ‘Lessons Learnt’ list features locking down PTP so only a few ports are actually allowed to provide time information to the network, dealing carefully with audio protocols like Dante which need PTP version 1 domains, and making sure all switches are PTP-aware.

The video finishes with Q&A after a quick summary of SMPTE RP 2059-15 which is aiming to standardise telemetry reporting on PTP and associated information. Questions from the audience include asking how easy it is to do inter-continental PTP, whether the internet is prone to asymmetrical paths and how to deal with PTP in the cloud.

Watch now!
Speakers

Bruce Devlin Bruce Devlin
Standards Vice President,
SMPTE
Gerard Phillips Gerard Phillips
Systems Engineer,
Arista
Richard Hoptroff Richard Hoptroff
Founder and CTO
Hoptroff London Ltd

Video: Esports for Broadcasters – Part III

In the last of three sessions on esports, the RTS Thames Valley looks at how vendors for the traditional sports market can adapt and serve this quickly growing market.

Guillaume Neveux from EVS sets the scene talking about the current viewing figures (44 million concurrent peak viewers for League of Legends) and revenue predictions of a 40% increase over the next three years. This is built on sponsorship and, like TV, this takes the form of ad insertion, and programme sponsorship (i.e. logo on screen) to name but two options. Esports has an advantage as they can control the whole world the sport takes place in. This means that advertising signs can be placed, live, on objects in the live stream which are seen by the viewers but not by the players, something which has been attempted in traditional sports but has yet to become common.

Guillaume also looks at how Twitch and YouTube Gaming work, commenting that one of their big differences from traditional sports is the chat room which scrolls next to the game itself. This lends a significant feeling of community to the game which is seldom replicated in traditional sports broadcasting. In general, esports is free to watch. Freemium subscriptions allow you to reduce the number of adverts seen and also improve the chat options.

The next part of the talk spotlights some of the roles unique to esports. The Caster is analogous to a commentator. They are there to weave a story, to explain what’s happening on screen and to add colour to the even by explaining more about what’s happening, about the people and about the game itself. Streamers are individuals who stream themselves playing computer games who, like YouTube personalities, can have extremely large audiences. An Observer is someone who moves around the game world but is invisible to the players, they are analogous to camera operators in that they can control their own view of the world and are also responsible for choosing which views from the players are seen. Essentially they are like a sub vision mixer feeding specific shots into the main programme as well as, in some circumstances, creating dedicated streams of shots for secondary streams. Graphics operators are just as important as in other types of programmes although aspect ratios are all the more tricky and this also involves integration into the game engines.

Guillaume also covers the equipment used by esports broadcasters. EVS is a premium brand with products honed to a very specific market. Guillaume explains that although the equipment may seem expensive, the efficiencies derived from buying equipment designed for your workflow a notable compared to creating similar workflows out of other equipment typically due to the added complexity, maintenance and workflow fit. At the end of the day, much of what traditional sports and esports needs is similar – slowmo, replays, graphics insertion – so only some modifications were needed to the EVS products to make them fit into the needed workflows.

Watch now!
Speakers

Guillaume Neveux Guillaume Neveux
Business Development Manager EMEA,
EVS

Video: What is esports? A crash course in modern esports broadcast

With an estimated global revenue of over USD1.1 billion1 and a global audience of almost half a billion people2, esports is a big industry and all accounts report it as growing. Although it sounds different, when you look behind the scenes, there’s actually lot of equipment and production that a broadcaster would recognise, as we showed in this behind the scenes footage that we featured in a previous article

Press play below as a taster before the main video to be a fly on the wall for five minutes as the tension mounts at this esports event final.

In this today’s talk from the Royal Television Society, Thames Valley, we’re introduced to esports from the bottom up: What it is, who does it and which companies are involved. I think esports is special in its ability to capture the interest of the broadcast industry, but exactly what it is and how it’s structured…few actually know. That’s all changing here, with Steven “Claw” Jalicy from ESL.

Steven explains that ESL is the largest company that runs tournaments and competitions outside of the games publishers. He explains that, unlike sports such as tennis, athletics and football which don’t have ‘owners’, all esports games have publishers who are able to control the way that gaming happens and have the ability to run tournaments themselves or, in effect, franchise this out around the world.

 
Steven takes us through the broadcast chain. Usually held in a stadium, OB kit and temporary set ups are nothing new to to the broadcast sports community. The first thing which is a change however, is ‘in-game’. There’s a lot more to covering esports than tennis in as much as for a tennis match you can turn up with some cameras and ball tracking kit and televise the games. Whilst doing it well is by no means trivial, with esports there are many more levels due to the fact that we have human players who are playing computer characters; to experience both the real and the in-game drama you need camera angles both in the real world and within the game. These in-game camera operators are call observers and just like real-life camera operators, their task is to capture all the action of the game. Sometimes this is done by following the players, sometimes by a birds-eye-view camera, depending on the game and, as ever, the publisher.

Naturally when you have a peak viewership of over a million people, streaming and live content distribution is really important. ESPN and, more recently, Eurosport have been airing esports so it’s important to realise that linear distribution is very much part of the mix for esports, it’s not an on-line only thing, though most of the numbers shared are the verified streaming numbers.

Steven talks about some of the challenges ESL faces in delivering the highest quality streams with so many tournaments happening and then moving to remote operation.

ESL prefers to build their own hardware for several reasons that Steven explains which include having the result fully-customisable and simplifying replacements. Similarly, ffmpeg and other open-source encoding is favoured for similar reasons.

The discussion finishes off with an extensive Q&A session including the ‘sanctity’ of the players’ equipment, the threshold for choosing to use vendor equipment (EVS vs Mediakind), transport over the internet and much more.

Watch now!
1Statista revenue report
2Statista eSports audience report
Speakers

Steven Jalicy Steven “Claw” Jalicy
Global Head of Streaming,
ESL Gaming

Video: Who is Watching? The Challenge of Digital TV Measurement

With the rapid increase in video-on-demand (VoD) viewing, over-the-top services such as Netflix and mobile TV, working out who watches TV – and when and where – has become a complicated business. Can the TV industry keep up with the changes and the ever-growing need to measure TV viewing habits across devices, platforms and other new ways to watch?

This panel from the RTS, includes Justin Sampson from BARB talking about how they’re capturing iPlayer views which are increasingly important, particularly has sometimes a TV programmes are available on OTT before linear transmission. But there is still work to do capturing views on Netflix, Youtube, Amazon and other services.

In the same vein, Rich Astley, Finecast CPO, pointed out there is a lot of advertisement viewing which is also uncaptured.

Sky Media’s MD John Lister gives his opinions including discussing the strong ability of linear TV to build brands which, clearly, is more important than individual advert spots.

Sarah Rose from Channel 4 discussed the continued importance of overnight ratings which are still highly correlated to having a hit on your hands.

Watch the whole thing to find out much more, hosted by Kate Bulkley

Speakers

Matt Hill Matt Hill
Research and Planning Director,
Thinkbox
Rich Astley Rich Astley
Chief Product Officer,
Finecast
John Litster John Litster
Managing Director,
Sky Media
Sarah Rose Sarah Rose
Chief Consumer & Strategy Officer,
Channel 4
Justin Sampson Justin Sampson
CEO,
Barb
Kate Bulkley Kate Bulkley
Moderator
Journalist