Services Global Viewpoint – November 2020

Every so often, the broadcast industry gets sideswiped by a massive change in technology. We had the introduction of SDI, then progression to MPEG, widescreen, HD, 4K, and now IP, all in the space of just thirty years. But throughout all these changes, one common denominator monopolizes our thinking, that is, timing and synchronization.

If we could start television all over again then we do things much differently. For example, light-field cameras completely dispense with video frames giving us an easier ride when migrating to asynchronous IP, and object and binaural audio would be a given. However, the need to maintain backwards compatibility for our viewers has demanded that we take small progressive steps in our adoption of new technologies.

More fundamentally, the same is true for broadcasters. Even greenfield sites wanting to take gigantic leaps of faith often find they have to make small intermediate transitions as new products are being developed. And engineers upgrading existing facilities always need to keep one eye on integration as well as backwards compatibility.

Commencing with the design of the timing plane is often where most engineers start. Our historical legacy of synchronous video and audio requires us to always think about how the timing will work and how it affects transmissions to give viewers the best quality of experience possible.

Even with IP we have to constantly consider timing as we are fundamentally distributing synchronous and time invariant audio and video samples over an asynchronous network.

Although there are many debates emerging about latency specifications, the critical aspect of latency is that it should be fixed and determinate. Variable latency helps nobody.

PTP v1 is nearly 20 years old. Version 2 was issued in 2008 and version 2.1 has just been released, demonstrating it has a great provenance (at least in modern technology timelines). Although PTP could be improved, I’m a great fan of it because it is stable, has a specification that is easily accessible, and is widely used in other industries.

To help migration to IP I think it’s important to separate our concept of time from that of latency and to think of broadcast television as a sequence of events. As I see it, time is an absolute sequential point when an event occurs with predictable measure, but latency is an inherent delay that cannot be avoided, but it can be and should be accurately calculable.

It’s a given that we must use credible time sources as this provides the source of truth for our time events. I’m more than open to considering other time systems but at this moment PTP is an industry standard that meets our required accuracy and has a specification that is easily accessible to anybody in the broadcast community.

In my view, what does have to change is our attitude to latency. If we start treating video and audio as predictable events, then we can accept latency and work with it within a fixed practical tolerance that can be embraced, and not a random variable to be feared.