SDI is a reliable and familiar technology, but it is time to migrate forward to IP-centric infrastructures and solutions.
As broadcasting hurtles at unprecedented speed to IP, exposing antiquated working practices, laying bare integration issues and questioning our very understanding of how television should work, are we a bit guilty of looking at SDI through rose-tinted spectacles?
In the 270Mbps days, SDI was new and difficult to make work. A T-piece stub or acute coax bend could cause all kinds of CRC and dropout issues. Large runs of cable had to be re-laid as incompatibility issues became apparent.
But to its credit, when SDI worked, it worked very well.
Just as we were getting to grips with standard definition, 1.485Gbps and even 2.97Gbps guised as 3G appeared. We were squeezing more and more out of the humble coaxial cable resulting in shorter cable runs and less resilience to interference.
Then we started talking about 4K and 8K needing as many as twelve, 3G coaxial cables per video chain. Imagine an SDI router with twelve coaxial cables per video channel. Logistically it’s really difficult to get that amount of connectivity density into the backplane of the router, or through a already full cable duct, but even more alarming is the amount of weight the coaxial cable and associated connectors and metal work needed to support this format.
Chroma Key Issues
SDI solved a specific problem, it could transport SD and HD baseband television signals within a building or outside broadcast vehicle with ease and reliability. But it did have its challenges. In the analogue days we would take RGB directly out of the camera for the chroma-key source, in effect providing a 4:4:4 system to give the chroma-key processor the best possible information to make a decent key. When we moved to SDI the color subsampling changed to 4:2:2, thus halving the color information for the keyer and making its job very difficult. It took some time for the manufacturers to overcome this challenge.
Flexibility is also an issue for SDI, there are many SMPTE formats, but they are very rigid and limited in number. This is both SDI’s greatest strength and Achilles Heel. Users are demanding more from broadcasters especially as we are constantly compared to the offerings of internet broadcasting.
One of the major benefits of internet broadcasting is the ability to receive information from the viewer and to take advantage of new and deep meta-data. Receiving feedback from the viewer helps fuel the immersive experience and keep them engaged. Broadcasters have more competition than ever from many diverse sources, not all of them from the tradition broadcast sector, and need to keep the viewer watching their channel.
Distributing meta-data in SDI is possible, but cumbersome and limited. However, with SMPTE2110, we will be able to distribute meta-data over IP and maintain the timing relationship to the original video. Click to enlarge.
Advertising companies now know how easy it is to get viewing data from set top boxes, so they can both align their advertising to personal needs, or know exactly what the viewingfigures are, instead of relying on the old extrapolated method of small viewing samples.
VANC is Full
Compared to IP, SDI makes handling of meta-data very difficult. It can be done, and is, but using VANC and data embedders and de-embedders is a difficult and involved task, requiring great skills and experience. Quite often, a simple stream of meta-data cannot be inserted into an SDI stream and must be sent by a convoluted route, thus negating the essence timing advantages SDI is soften hailed for.
Distributing meta-data in IP is easy using software development tool kits (SDK’s). We have the issue of timing and keeping the meta-data synchronized with the associated video and audio essence, but as SMPTE ST2110 gains popularity, the use of Precision Time Protocol (PTP) specifically in ST2110-10 will make this easier and open development to a wider pool of innovators.
8K120P-444 needs 120Gbps
High Dynamic Range is gaining focus and systems such as Dolby Vision’s PQ system work better with meta-data. This can be facilitated in SDI but will be much easier with IP. Furthermore, software developers will treat the meta-data as a simple data stream and be able to process and log it in standard servers without having to worry about esoteric hardware devices such as SDI de-embedders, thus making it a candidate for virtualization and cloud processing.
To suggest IP makes distribution of 4K and 8K easier is a bit of a cheat. 8K120P-444 needs connection speeds of 120Gbps, well beyond what is available now in the IT world. However, IT networks are only getting faster and the big corporates such as Cisco have development budgets that would make the average broadcaster manufacturer cry, and these speeds will come soon.
Applying light compression using standards such as TICO will ensure 4K and 8K distribution will fit down a 40Gbps network. But there is a lot of concern about compression, many traditionalists maintain we must distribute a signal as base-band, but the definition of base-band lacks clarity. As we have seen, when we moved from RGB distribution to SDI we halved the color bandwidth, so is this still base-band?
Frame Sampling is Compression
And frame sampling is also a form of compression. We’re taking an infinitely varying scene and sampling it at 25fps or 29.97fps, mathematically we’re throwing nearly all the data away. So, if we’re worried about compression, surely, we should be more worried about video sampling at the frame rates we use?
To transport six HD uncompressed signals requires on single fiber with diameter of 3mm, or six SDI cables with a combined density of 30mm, not to mention the weight differences. Click to enlarge.
SDI is always hailed as incredibly secure because the only way to hack into it is to use a pair of wire cutters, but this assumes a system that has no IP control or configuration. Most modern SDI processing equipment, such as transcoders or even distribution amplifiers, have an Ethernet/IP connection on them connected to a network. At this point, a potential hacker has access into the SDI system. Not all broadcast engineers are that diligent with their logon credentials, and user names of “admin” and passwords of “1234” are quite common in broadcast kit. A hacker dream. It only takes one person to click on a phishing email to take down all your SDI kit.
We are moving to an IP world, and with good technical justification and reasoning. Furthermore, it just happens to be supported by an outstanding business case which can only be for the benefit of the viewer and broadcaster. It’s easy to look at the good ‘old days and reminisce about how wonderful SDI was, but IP will also be wonderful and provide us with a better viewing experience as technology and innovation develops around it.
Tony Orme has written a series of tutorial articles on moving to IP centric software and hardware infrastructures. To learn more about migrating to this new platform, type "Tony Orme" in the search box on The Broadcast Bridge home page. You will be provided a list of his articles and links to each of them.
Related Editorial Content
For many years users have been reluctant to pay for software maintenance, especially when they include bug fixes. But is this a bit unfair to the vendors? A plethora of media file specifications and formats are available to broadcasters, from…
The business case for migrating to IP is compelling and driven by the needs of business owners. Broadcast engineers must rise to the challenge and if they are to deliver reliable IP infrastructures they must understand not only the technology,…
IT engineers use generic propriety models when designing networks. As television moves towards IP, broadcast engineers must understand these networks and how they work. This series of articles explains networks from a broadcast engineers point of view so they can…