This multi-post series is a dive deep into the Society of Motion Picture and Television Engineers (SMPTE) ST 2110 suite of standards. The series explores the reasons customers transition from Serial Digital Interface (SDI) infrastructure to IP. It also looks at the trade-offs to consider when making the switch, and best practices to use AWS Elemental Live with SMPTE ST 2110.
Part 1: Background and key benefits of SMPTE ST 2110 on AWS Elemental Live (this post)
Part 2: Working with 2110 on the AWS Elemental Live Appliance
Part 3: Simplifying with NMOS
For years, traditional broadcasters relied on SDI signals to transmit uncompressed video and audio signals between devices. SDI transmits a muxed feed of video/audio/ancillary data unidirectionally. The cost of SDI infrastructure remains high, as the demand for broadcast solutions is not as wide as the professional-grade IT market in general. SMPTE ST 2110 is primarily for live studio use cases to improve on the foundation that SDI laid. Its counterpart, SMPTE 2022-6 is designed to allow uncompressed contribution over IP, and showcases a method of encapsulating SDI for IP transport called High Bitrate Media Transport (HBRMT). 2022-6 is solely focused on getting SDI over IP. Where 2110 really stands apart is the ability to transmit isolated essence flows. For example, a 2110-30 feed only transmits the uncompressed PCM stereo audio portion of the media at ~3 Mbps. It does not contain video or ancillary data, which would add another ~3 Gbps of unnecessary overhead. This reduces the bitrate of the flow by 99.9%. The 2110 standard is built on many well-known standards from providers such as IETF, ITU-R, AES and IEEE, relying on proven technologies to ensure its success. Elemental Live appliances support 2110 inbound and outbound, as customers use them in a plethora of ways to support their unique workflows.
A major key benefit of 2110 is the transmission of the essence flows. It removes the need to transmit and demux the other components not required for that portion of processing, which would add unnecessary latency. For example, an audio mixer shouldn’t spend processing cycles demuxing video and audio just to throw away all the video packets. It greatly reduces the bandwidth requirements. Because these essence flows can be made available in a multicast network, multiple systems can receive and process the flows independently from each other. This eliminates the need for any additional signaling or cabling, which makes redundancy much easier. This also improves scalability and agility by allowing you to hand off signals to more receivers without adding additional cabling.
Let’s compare a simple example of a single Ultra-High Definition (UHD) feed that uses 3G-SDI vs 2110. The obvious area to simplify is the cabling. To get the 12G source for a UHD feed into Elemental Live, four 3G-SDI cables are used (8 if you have a hot backup). Doing so uses up a significant portion of your video router’s output capacity. Since SDI is unidirectional, there is no return feed capability from the appliance. This means that all outputs are IP-based, so engineers need to maintain multiple network stacks to process the video. Since Over-the-Top (OTT) delivery is based on IP infrastructure, the consolidation is beneficial for many broadcasters as they look to expand to new markets. With 2110, the inputs and outputs can use the bidirectional NICs to greatly simplify the cabling requirements. The network hardware itself is cheaper, less specialized than broadcast-specific gear, and is better understood by general system administration. Furthermore, if more signals are added to the workflow, it’s unlikely new cabling is needed. This makes 2110 more agile, scalable, and often more cost effective. For example, the newest generation of Elemental Live appliances can ship with two 25 GbE NICs. They can support a large number of 2110 flows, both inbound and outbound.
The SDI standard changes slowly (HD-SDI, 3G-SDI, 12G-SDI), but the need to support greater resolutions continues to move swiftly. Often when the upgrade in SDI arrives, it means a large capital expenditure and renders earlier broadcast stacks useless. It’s far more likely that a simple hardware upgrade to IP infrastructure allows access to the newest resolutions and workflows when using 2110. Doing so helps keep businesses future-proof and keeps costs down. This assumes that the IP network switch has the ability to handle high bandwidth through its backplane.
SMPTE ST 2110 is broken down into a series of documents; each tackles one specific part of the standard suite. Including explanations of how flows are synchronized using the Time Protocol (PTP), best practices for encapsulation of video, audio, or ancillary data using RTP inside UDP inside IP packets. Restrictions around packet size and how to ensure that 2110 senders and receivers are in compliance by properly configuring the “session description” of each flow using the Session Description Protocol (SDP).
The following is a breakdown for reference:
2110-0: Overview of the 2110 Standard
2110-10: System Timing and Definitions
2110-20: Uncompressed Active Video
2110-21: Traffic Shaping and Delivery Timing
2110-30: PCM Digital Audio
2110-31: AES Transparent Transport
2110-40: Transport of SMPTE Ancillary Data
DURING THE TRANSITION
Nothing happens overnight in broadcast, so to help bridge the gap between SDI and IP, there are encapsulators that can convert sources between these two types. SDI can be split into multiple 2110 essence flows, or multiple 2110 flows can be muxed back into a single SDI feed. This ensures that legacy equipment is used until it is replaced by newer models that support 2110. A great example of this is when major broadcasters and content providers work with affiliate stations. Not all affiliates can upgrade from SDI at the same time, so this provides a stop-gap solution in the interim.
It’s also good to start exploring PTP grandmaster clocks (aka reference clocks) as soon as possible, as its a critical part of the timing for 2110. Network Time Protocol (NTP) is simply not precise enough to resync the essence flows properly. NTP can be accurate to about 1 millisecond of UTC time, whereas PTP can be accurate to the nanosecond. As more manufacturers start to integrate 2110, more clocks have added PTP functionality for use in broadcast data centers. They often pull their time via GPS satellites carrying atomic clocks for absolute precision. It’s worth noting that Live does not consume the reference clock directly with 2110 as it does with SDI inputs. Instead, it must be embedded as ancillary data.
There are two main points to consider with a shift to 2110: monitoring and security. For decades, distribution amplifiers or SDI video routers have been used to duplicate an SDI feed, enabling multi-viewers, eyes-on-glass monitoring, and third-party QA appliances to ensure signals look and sound correct. The SDI signal is always the same, heading in a single direction with a fixed number of embedded audio channels. There is a very simple test to see if the source is accurate – plug it in and it shows up. That is not the case with 2110, so it is treated differently. For example, 20 of the 2110 flows head one direction and 30 more head the opposite all on the same cable. IP network monitoring tools must “see” the signals, and ensure the proper shaping of the traffic so it doesn’t overload the network switch. Spine-leaf network architectures help avoid such overloads. Specialized hardware is available if needed to do more advanced analysis and monitoring of multiple 2110 signals. However, there is a substantial cost factor to consider.
With regards to security, I have not heard of an SDI feed being hacked. That’s not true with IP, so guarding your IP video network is much more critical. Firewalls should be put in place, your Elemental Live appliances should never be publicly accessible, and they should always run the most up-to-date software.
To summarize, SMPTE ST 2110 is agile because it isolates the essence flows and scales. It does so by relying on IP infrastructure that does not need multiple network stacks. As more studio and broadcast workflows move to the cloud, it is the next phase of evolution to allow seamless content sharing and modular processing of the essence flows to build the next generation of media workflows. In the next post in this series, I look at how to use 2110 on AWS Elemental Live.