As the broadcast industry migrates from SDI-based to IP-based infrastructures, the switching of media streams in IP production workflows, and the standards associated with video transport over IP networks, add a level of complexity for the broadcast environment. This article summarizes methods for IP media switching and the associated standards involved.
Switching of media streams in an IP-based production setting requires seamless, known as glitch-free, switching of media streams between sources. A media stream refers to a video stream, audio stream, or ancillary data stream. Video streams are the focus of this article because they require significantly more bandwidth than other media streams. For comparison, a single uncompressed HD video stream requires over 700 times as much bandwidth as an uncompressed stereo audio stream. Video streams need to be switched at frame boundaries. Additionally, switching between video sources may involve visual effects that provide a seamless transition (for example, cross-fading, image dissolving, etc.).
Switching of media streams in SDI-based workflows is a relatively simple process:
- SDI streams in the production environment are synchronized to a common reference (black burst or tri-sync).
- A switch point is identified using a SMPTE Recommended Practice (RP168).
- The stream is switched at the identified switch point.
As broadcast networks migrate from SDI to IP, they can choose from several methods for switching media streams. In the simplest method, switching can happen after the streams are de-packetized, which means the media payload is removed from the IP packets to allow switching to be handled in a similar manner to SDI. Although this method is conceptually simple, it is the least flexible in terms of networking architecture since all media streams need to be switched at endpoints.
A second method involves switching media streams in the IP domain, which means switching before the media streams are extracted from IP packets. This method takes advantage of the flexibility of modern IP networks that allow traffic to be routed to multiple locations under software control through distributed switching. But it also has limitations, addressed below.
Emerging studio IP transport protocols (like the draft ST2110 standards) enable media stream switching in the IP domain outlined in the second method above. ST2110 includes the ST2059 timing protocols that realign media streams that have taken different paths through the network. These timing protocols use algorithmic methods for the realignment instead of depending on the matched point-to-point connections used in SDI transport. This allows for greater flexibility in the routing and switching of media streams, but it also adds complexity to switching within IP networks.
Despite multiple options for IP switching, the current technology does not allow streams to be seamlessly switched within IP networks. In order for a seamless switch to be performed, both streams need to be delivered to a media endpoint, which performs the switch function. In this scenario, the overlap period of the media streams may be quite short – just enough time for the endpoint to perform the seamless switch. Afterward, the network connection may be freed up for other traffic as needed.
If the switchover is a simple seamless switch (that is, it does not include visual effects), the endpoint can align the two streams and switch between streams at the video frame boundary. It is not necessary to extract the media from the IP streams before the switchover. This can reduce switchover time and complexity in the endpoint. If a switchover with visual effects is required, the media streams will need to be extracted and processed in software.
An option for distributed seamless switching is to create a “restricted” endpoint – that is, create an endpoint whose only role is to switch between media streams, not to extract active media streams to media interfaces or processing elements. Customized switch features could be implemented to perform this function. Distributed switching makes IP networks flexible for routing media streams.
Open standards such as the Networked Media Open Specification (NMOS) enable media stream routing, switching, and processing to be distributed through a broadcast production data center that includes equipment and software from multiple vendors. NMOS and associated standards, as well as the emerging IP transport standards, will ultimately enable fully virtualized data centers to be leveraged for live media production. Fully virtualized live media production will enable traditional and non-traditional broadcasters to provide live event production, bringing high-quality coverage of news and sporting events from all over the world to global audiences.
The transition to IP-based workflows for live production is a pivotal change for the broadcast industry. Flexible switching and routing architectures for high-bandwidth media streams are a critical part of that transition. Despite the complexity of IP-based workflows, progress towards standards-based solutions is well underway.
Marc Levy is chief technology officer of Macnica, a provider of products that implement the industry-defined standards for moving live, high-definition video over standard IP (Ethernet) networks such as SMPTE2022-1/2/5/6/7 and TR-03 (SMPTE2110).