A Utah Scientific White Paper

The Pillars of the New SMPTE 2110 Standard

By Scott Barella, CTO of Utah Scientific

Introduction
I remember, years ago, looking through the SMPTE 259M standard for the Serial Digital Interface (SDI).
It was a bit much to grasp at first, but we had plenty of help from industry veterans as we began to
navigate what is now a very common digital video standard. Needless to say, SMPTE 259M was
revolutionary for its day. The SDI standard replaced many vital pieces of analog video such as four-
field color framing, vertical interval data, sync, breezeway, color burst and a slew of details that helped
engineers understand a very mature video format on coaxial cable. It would take me a while before I
could manipulate SDI like I did with the old analog signal.

When SMPTE 292M came along, it more or less followed in the footsteps of the SMPTE 259M standard
— and before long, we were quickly adapting studio video to high definition. One of the “comfort
foods” of that era was that these changes were all happening on coaxial cable, the same cable that I
made connectors for when I was a junior television engineer. Fast forward to the present, where we’re
working with high-speed fiber at rates that we never would have believed back in the old analog days.
And, like all good things, coax will one day be completely replaced by fiber. Now, that’s a mouthful,
because it’s not just a little cable switch — it’s a huge change for our industry. Like any technology
change, coax signals will be adapted and moved into a technology sector we’ve all seen growing
alongside studio video: Ethernet. But since there have been a few unsuccessful attempts at moving
the SDI signal to Ethernet, it’s important to understand how we arrived at the present state of SDI in
the industry.

March 2017 1
The evolution of a standard

A number of years ago, the SMPTE 2022 committee developed a method to use packets for
compressed MPEG transport streams, and in the process gave the industry the ability to use
compressed video and audio and move them onto Ethernet switched circuits, in much the same way
as we used ASI over switched circuits. As ASI payloads were usually well under 200 Mb/s, with little
bandwidth usage, this seemed like it would be an easy transition.

But truthfully, we learned a ton of information about moving video using 2022, and it inspired some
people to think about how to apply the standard in the uncompressed domain where the HD payloads
were higher. Actually, a lot higher – more like 1.5 Gb/s. In the beginning, the task seemed pretty
daunting considering that 1-Gb/s port speeds were common, but then 10-Gb/s ports started to
appear, making it easier to aggregate switches. In other words, these 10-Gb/s ports were primarily
used to daisy-chain 1-Gb/s ports and switch them together to address non-blocking architectures and
therefore avoid packet collisions, or an overrun of the switch backplane. Before long, the first 10-Gb/s
Ethernet switch appeared, and engineers who were looking at the next frontier thought that this might
be the way to move uncompressed HD video.

There were some early attempts at standardizing this approach from groups like the IEEE, whose
Audio Video Bridging (AVB) scheme (IEEE 802.1BA) was a good attempt at addressing the needs of
professional uncompressed video. AVB had a very interesting synchronization scheme that was quite
similar to the SDI sync used in the SMPTE SDI signal. But one of the main requisite AVB components is
a specialized Ethernet switch. The format worked nicely, but the switch requirement was a problem
for some who thought the uncompressed video signal would be better adapted to Ethernet standards
and use commercial off-the-shelf (COTS) Ethernet switches. In other words, many felt that the signal
needed to fit Ethernet rather than making Ethernet fit the uncompressed video. The debate is
interesting, but in the end, COTS switches would be the reason for the demise of AVB. Meanwhile, the
quest continued for a Studio-Video-over-IP (SVIP) solution.

While the AVB camp was busy, so too were engineers who thought that video could be encapsulated
in the same manner as 2022-1/2— so the SMPTE committee drafted and later ratified SMPTE 2022-
5/6/7. SMPTE 2022-6 laid out a straightforward method for encapsulating SDI video that worked nicely
for single-ended program contribution, in which the focus was to deliver a complete, uncompressed
audio and video stream without too much need to take it all apart, manipulate it, and put it back
together again. While this was a pretty slick method to make uncompressed video, it wasn’t
necessarily a replacement for SDI because it would force operators to un-encapsulate the audio and
video and then de-embed the audio. This was necessary for manipulating the signal to perform audio
and video chores such as squeezing, adjusting levels, mixing, etc.

March 2017 2
The Joint Task Force for Networked Media (JT-NM) was a consortium of SMPTE, EBU, and VSF that put
together a roadmap for SVIP, with the express purpose of defining an industry standard for
uncompressed IP video and audio. While other companies were working on a proprietary means to
help their customers, the work of JT-NM had an industry-wide focus. Finally, in 2014, the VSF published
Technical Recommendations 03 and 04 (TR-03/04), which address the shortcomings of SMPTE 2022-6
for using the signal as a replacement for SDI. Key to these recommendations is the idea that signals
are separated as pure video, audio, and data streams, making it much easier to manipulate each of
the three elements. At the heart of this idea is the use of Precision Time Protocol (PTP) as described in
SMPTE 2059 -1/2. PTP is used to synchronize the three signals together using time stamps on the RTP
packets that encapsulate each signal. In this manner, each stream has a relation in time to the others,
therefore solving the lip-sync issues that have plagued our industry for years. In this manner, all signal
streams including genlock would be carried over one cable alongside others in a 10-Gb/s port.

Enter SMPTE 2110

Shortly thereafter, a consortium of users and vendors calling itself the Alliance for IP Media Solutions
(AIMS) came together to implement TR-03/04. AIMS would later adopt the work of the SMPTE 2110
Draft Group, which was working on an industry standard based on these recommendation
documents.

The SMPTE ST 2110 standard includes several sections:

! 2110-10 describes the system timing and how RTP packets will be used, as well as how each of
the streams will be carried in the network. This section, more or less describes how SMPTE
2059-1/2 should be used and the PTP packets are used as reference when “stamping” video,
audio, and data packets.

SMPTE 2059-1/2 is a separate standard that, by nature, creates a new genlock for IP that is
similar to the black reference and tri-level sync of SDI. The difference is that this signal is not
carried by a separate cable as in the past; instead, 2059 allows the video, audio, and data to
be stamped by PTP packets within the network. Giving a single fiber the ability to carry not
only video, audio, and data but also genlock is a huge development, giving broadcast
engineers a powerful new ability to address lip synchronization errors.

The key to SMPTE 2110 is this ability to time stamp each of these three elements. When each
has a time stamp, they can be used to compare other signals against the time marks that are
in each separate stream. Not only can they be used for timing the composite streams for
switching or other “timed” events, but they can also be used within the streams such as the
relationship of video to audio and video to data in the case of closed captioning. Our industry
has never had the ability to mark the video, audio, and data and finally has a nice tool to solve
the problems of lip synchronization!

March 2017 3
 ! 2110-20 video uses the internet IETF standard RFC 4175 as a sort of template. The main
feature of this standard eliminates the need for the vertical blanking interval and other
historical encapsulations of additional signals. It also provides a means to incorporate just the
video essence that creates lines of video, no matter the desired resolution and frame rate. In
other words, the video supports the fundamental pixel video stream without the burden of
sync and Video Ancillary Data (VANC); just the pixels that make up lines of video.

! 2110-21 addresses the timing models for video and addresses present FPGA designs as well
as designs that will be software-based. These include a timing model for “narrow” timing for
much tighter specifications, as well as a model for “wide” timing for software models that have
not yet reached the kind of precision that dedicated hardware can offer. In other words, these
two-timing models offer users some flexibility for current designs based on hardware and
future implementations using software designs.

! 2110-30 leverages AES 67 for the carriage of uncompressed PCM audio signals, allowing the IP
audio signal stream to be transported separately and thereby eliminating the need to
“unbind” the audio from the video; i.e. discreet or unembedded audio. It took years for audio
engineers to come to agreement on AES 67, a single uncompressed standard for audio not
just in paired channels but in multiple channels of audio – creating the possibility of carrying
hundreds of channels.

! 2110-31 addresses legacy AES3 audio. The industry has been using this format for the past
couple of decades and will certainly require the ability to support legacy metadata formats
into the future, as well as the essence audio of this format.

! 2110-40 addresses ancillary data that was once inside the SDI VANC area of the vertical
interval, but in the new IP environment will be handled as a separate RTP essence that is not
bound to video in the manner of current VANC data. CEA-608 and CEA-708 closed captioning,
timecode, AFD, and other VANC data will be separated.

! 2110-50 comes from the VSF’s TR-04 and is a derivation of 2022-6 and AES 67, using SMPTE
2059-1/2 as the timing mechanism. This may be of particular interest to operators who like
the simplicity of 2022-6 but need to provide separate AES 67 audio.

March 2017 4
So where do we go from here?
As IP interoperability standards like SMPTE 2110 continue to mature, they’ll pave the way for
broadcasters’ eventual wholesale migration to all-IP operations. In the meantime, many operators are
looking at an incremental approach for migrating a baseband signal plant to IP. Because SDI routers
remain at the very heart of most broadcast operations – in fact, all of a current plant’s signal types go
through them – they provide an ideal place for exchanging signals. By leveraging the power of SDI
routing, both IP and SDI environments can coexist in this hybrid fashion for years to come. If history’s
a guide, broadcast equipment will begin to ship with both SDI and IP ports in further support of this
hybrid strategy. And that’s where Utah Scientific comes in.

With a 40-plus-year track record of providing leading-edge broadcast routing technology, we’re
committed to supporting our customers through every technology transition, especially those that are
fundamental to core operations. One of our newest innovations is the UTAH-400 IP Gateway family of
input and output cards that provide two-way conversion of SDI video signals and SMPTE-2022 signals
over a 10-G Ethernet connection. In most broadcast infrastructures, legacy SDI video devices aren’t
going away anytime soon – and these new IP gateway cards help to bridge the gap between traditional
SDI and IP signal processing.

March 2017 5