The TV technology industry needs an IP reality check

Schematic IP production workflow - inverted image
Peter Schut, Axon
November 4th 2016 at 9:29AM : By

Before the TV industry blindly stampedes into the all-IP future, Peter Schut, CTO of Axon, thinks it’s time for a reality check

Whilst the transition to IP is inevitable, it is clear that the road ahead is not without its pitfalls and distractions. Customers and manufacturers alike face a steep learning curve as we move into an IT-centric environment, addressing the challenges of different applications and format limitations head on. For those trying to navigate their way through this shifting landscape, it’s time for a reality check.

IP is not the end goal

At the time of writing this article, it could be argued that today the barriers to IP outweigh the opportunities. Currently, IP is less reliable than SDI, more expensive and harder to control. In its basic format, Ethernet does not work in a broadcast environment as it is built around a best-effort delivery strategy. However, there is no doubt that it will become the backbone of our industry: with protocol enhancements, it will prove to be more convenient, more flexible and eventually cheaper.

However, what we should not forget is that IP in itself (or Ethernet, a safer name as Layer 2 is not yet dead) is not the final goal – it is a path to the next stage. As an industry, we must prepare for a future where data centers will manage the vast amount of the processing we will need to perform, and these may not necessarily be located close to the action. These data centers will be limited to Ethernet I/O and this makes the future backbone of our industry Ethernet-centric. It is this end goal we need to keep front of mind.

Our industry has a tendency to carry old formats for decades

One size does not fit all

Whilst our future lies in Ethernet, what is becoming increasingly clear is that every application needs its own specific technology – one standard does not fit all - and this is leading to confusion amongst broadcasters and manufacturers alike.

Taking a snapshot today, it looks like TR03 will become the production standard with separate elementary streams for video, audio and data. It will be a timed and low latency format. ST2022 is most definitely a contribution standard. It has all data in one stream - audio, video and data locked together in one. ST2022 is nothing more than packetised SDI with embedded audio and data.

This is good for contribution applications where you do not want to mess with individual payloads. Playout, however, is a completely different game, supported by different formats - from SDI, uncompressed IP to files. The result is transport streams with highly compressed data payloads. They could live in an ASI stream and move increasingly into Ethernet all the way to your home. With broadcasters staggering rollout according to different operational priorities, expectations may have to be reset.

Elvis (SDI) has not left the building

How does a broadcaster start to plan the move from SDI to IP? Let’s be realistic here. Our industry has a tendency to carry old formats for decades (we still sell composite video products) and the move to IP will not change that. SDI is not dead, despite the marketing hype. It may now be viewed by many as a legacy format, but SDI remains a mature and reliable interface that our customers use and trust. In all honesty, today there are many applications where the use of IP is questionable and SDI is the only route forward.

Over the next 5 years we will slowly move into the IP domain, but for today we must navigate a hybrid world with its center of gravity in SDI. Whilst SDI and Ethernet will live side by side for the foreseeable future, Axon is providing the best interim solution allowing companies to take a pragmatic approach to deliver the efficiencies of IP with SDI as a sticky plaster to protect operations from its limitations.

Early adopters’ pain is our gain

Every step-change relies on early adopters, but those broadcasters brave enough to lead the charge towards IP face the same problem: a lack of native IP products. Whilst their new infrastructure cores are based on IP, the surrounding environment is still dominated by SDI and some are now reintroducing SDI routing because IP routing is simply not maturing fast enough.

The great thing about these early adopters is the experience they provide to our industry, to both manufacturers and users and for that we should be grateful. Somebody has to start and they are doing it. They are dealing first-hand with latency and control issues, bullying along the major IP switch manufacturers as they struggle to overcome the challenges of delivering a completely different approach to IP routing. Resending a packet, something very normal in the IP/IT industry, is not something a live production facility can afford.  Their pain now will hopefully be our gain in the future.

Currently, IP is less reliable than SDI, more expensive and harder to control

The battle for open standards

Undoubtedly to move forward we need a commitment to interoperability and open standards - preferably a few standards to keep us in business but not so many that it becomes too expensive to be commercially viable. Whilst the debate on standards ensues and new developments emerge with increasing frequency, manufacturers must perform a balancing act, choosing to respond or bide their time.

Support for ST2022 is a case in point - the first step taken in the IP domain of any serious magnitude. As previously mentioned, this standard is designed for contribution/long haul applications. Whilst Axon has worked closely with the EBU, VRT and industry partners to deploy 2022 in such initiatives as the LiveIP project, from the outset we saw the limitations of the standard in live production applications. Whilst we have championed AVB (Audio Video Bridging) as an effective alternative, the industry continued to focus on ST2022 for live applications, turning a blind eye to its limitations and more or less neglecting its drawbacks.

Now the blinkers are off and TR03 is the new kid on the block, with a technical recommendation in draft to become ST2110. I believe TR03/ST2110 will be a much better standard for live applications, but there is still some way to go. AMWA is using TR03 as the core for what they call the Media Network Incubator (MNI) - not to be confused with Sony’s NMI initiative. Their developers are working on a practical implementation of TR03, to add control in the form of NMOS and to provide an environment that allows interoperability tests and establishes best practice.

AIMS on the other hand is an organisation that wants to make sure our industry is backing a single horse instead of hedging our bets on every horse in the field. With over 50 members including Axon and other major players, AIMS has become the body it was intended to be, driven by a clear mission to eliminate fragmentation and maximise hardware and software interoperability - a goal Axon, as an open platform provider, fully endorses. 

Where now for Axon?

Over the last 12 months, Axon has been working to develop a clear path for our customers; we have moved from AVB, via ST2022 to TR03 (and TR04). This shift in focus in itself is not a big deal – our development team has always been well placed to react to the market - but these technologies are as equal as they are different and all are very complex. It is clear that ST2022 is here to stay in some applications. Whether TR03 will be embraced completely by our industry still needs to be seen. The signs are all there. When that happens, we will have to deal with both 2022 and TR03 for many years to come.

Axon’s strategy remains both customer-focused and pragmatic. We are committed to delivering products and hybrid solutions that the industry needs as it moves from SDI to IP, providing control for these products with our latest NMOS-ready Cerebrum software platform. In addition, Axon remains agnostic, working on the next generation of products where Ethernet is the default interface and SDI is finally put to rest with the respect it deserves.

Though we all have hope for the future, currently IP is less reliable than SDI, more expensive and harder to control.