Several sessions and many companies talked about when, if and how the broadcast production facility will move away from SDI-based video workflow connectivity to one based on using IP protocols.
A paper presented by Peter Chave of Cisco nicely summed up the current situation in broadcasting. SDI is still the predominate transport medium within capture and production facilities, but it is becoming less predominant for the movement of video between facilities. Here, IP distribution already has a strong presence in the distribution part of the chain and in some ingest operations.
Chave said that broadcast industry faces a decision point as it moves to building infrastructure that can handle 4K-UHD content. Quad-SDI is not a rational solution as it requires 4 times the cabling, adding significant cost, and it will be hard to synchronize. Another option is to develop 12G-SDI coax cable (SMPTE 2082), but the cost and limited distance potential with this cable seems questionable too (see Is a Coax 4K Solution Dead on Arrival in Broadcast? http://tinyurl.com/ngry4c7). A third option is to consider some mezzanine compression and use the existing HD- and 3G-SDI infrastructure, but will this cause artifacts and unacceptable latency?
There are clear forces that are moving the broadcast industry in the IP direction. This includes existing and pending standards suited for the broadcast industry; the “virtualization” of operations and functions to the cloud, which is IP based, of course; and the push toward UHD, which will require a technology refresh.
So in the move to an all IP infrastructure, what are the key questions? Chave summed them up as:
* How to encapsulate and transport UHD live signals
* Minimizing delays/latency
* Flow control, congestion control and quality of service (QoS)
* Reference clock/video synchronization
Some SMPTE standards exist to support this transition to IP transport including SMPTE 2022-5, and -6 which provide for forward error correction (to recover lost packets) for video up to 1080p/60. There are no standards for UHD transport.
Another big challenge is the synchronization of video. This is being addressed by adopting an SMPTE version of the IEEE 1588 standard called Precision Time Protocol (PTP). This standard specifies a grandmaster clock and a start time (epoch) from which everything is referenced. However, each node in the IP network must support PTP to transfer video with proper synchronization. SMPTE has a working group that is now developing a “profile” and SMPTE standard that suits the needs of the broadcast industry for IP distribution of video. This will become the SMPTE 2059 standard and is nearly ready for final approval. This will enable IP timing synchronization within one subnet (layer 2 networks) or between networks (layer 3 network architectures).
While SMPTE is developing its own standards for an IP infrastructure, there is another effort underway called Audio Video Bridging (AVB) (see article: Is AVB Coming to Broadcast?). This standard also uses IEEE 1588 and meets the requirements laid out by Chave above, but it is currently on a Layer 2 protocol which limits its functionality for network to network transfers. However, a Layer 3 standard is in development and could well suit the needs of broadcasters in the future.
However, Chave thinks AVB is behind SMPTE’s efforts. SMPTE does not seem to be doing too much in the areas of standardizing Chave’s third bullet above (Flow control, congestion control and quality of service (QoS)) because he feels proper set up of the network can handle these requirements.
Chave said Cisco has been demonstrating the ability to send 4K video over standard IP networks for over a year when he first showed a video over IP solution using standard Ethernet switches in the 4K Demo room that Insight Media organized at the 2013 SMPTE Fall conference.
At CCW, he showed an updated version of this demo. 4K/30 content originated from an AJA KiPro Quad device and output over 4 SDI cables goes to the DCM-G gateway as uncompressed video. The DCM-G is used to encapsulate the 4K image into an IP stream. This is output over two 10 Gbps Ethernet ports (uncompressed 4K requires close to 12 Gbps so two 10 Gb channels needed) to a Nexus 3000 switch – a standard IP network switch product used in countless datacenters and costing less than $10K. The signal is then combined and sent out over a single 40 Gbps loop back into the Nexus 3000. Dual 10 Gbps links go back to DCM-G where it is converted back to SDI for delivery to the monitor.
As was noted by Chave and echoed by several speakers, we still have a long way to go before there is an end-to-end IP workflow starting with a single Ethernet bi-directional connection to the camera that delivers, video, genlock, metadata, tally and more and ends with an Ethernet connection to a consumer or professional display.
A session at CCW called “The IP-Based Facility: Is it for You? If so, When?” featured three presentations and a panel discussion. The speakers were from PBS, CNBC and CNN and each described how they were looking at the use of IP within their domains.
At PBS, the company is now in the process of developing its sixth generation content distribution system, which will now also feature integration with National Public Radio (NPR). Mario Vecchi, the CTO at PBS noted that the cost of a distribution system is always a factor, but it is not the driver – staying competitive in business is the real imperative, he said.
Others in the session agreed with this, noting that the move to IP is going to happen but it should not be billed as a cost saving measure. In fact, they said they would be happy if they were able to do it on a cost neutral basis. They noted, for example, that the move to IP will create savings in cabling and equipment, but this may be offset by increased network security and other related equipment costs.
Vecchi listed a number of advantages to moving to IP including better interoperability, universal connectivity and the ability to move from hardware based solutions to cloud-based software solutions (virtualization). The move also leverages the huge base of investment, development and equipment in the much larger IT industry and has the ability to move the cost of infrastructure development from Capex (Capital Expenditure) to Opex (Operating Expenditure).
PBS is rolling out a limited version of its new architecture in 2014 that will include about 18 end points and two master control rooms. He was also clear to note that this will not be done on the public Internet, but on private networks where it can manage the traffic. Its system features an MPLS (multi-protocol label switching) architecture, which is a scalable, protocol-independent transport mechanism. It is a layer 2.5 type system in that it can handle unified switching for both circuit-based and packet-based clients.
But he has concerns such as reliability, availability, security, troubleshooting and capacity management. Vendors will need to step up their investments and lots of testing will be needed, he thinks.
Michael Koetter, SVP for Media Technology and development at Turner Broadcasting, the owners of CNN, cited three specific reasons to consider bringing IP in the production environment: cable consolidation; simplified integration; wide area operation.
But there are issues: Training broadcast engineers in IT principles is a huge issue as there are no schools doing that today. The company has resorted to hiring IT professionals and teaching them about broadcasting. He also mentioned troubleshooting, standards and security concerns as impediments to moving forward. He noted that some companies are developing their own systems and these may not interoperate in a “mix and match” world, which will be an issue.
In terms of implementation, Koetter said that they are planning to move to a new facility called Greenfield in late 2018 and he wants to make it an all-IP based facility, but he is not sure if the industry will be ready to support that in time.
In the panel part of the session, moderator Robert Seidel, the VP of Engineering and Advanced Technology at CBS, asked if reliability in video over IP was robust enough to enable live video yet.
Koetter noted that SDI “just works” and IP has a long way to go to get to that point. Steve Fastook, SVP of Technical and Commercial Operations at CNBC said it remains high risk today and he likes to be a late adopter, so he will deploy only in low risk areas. PBS’s Vecchi said the company doesn’t have big real-time needs, so it should be OK with using IP for its distribution part.
One participant also asked if the move to IP will reduce the use of satellite-based transport. All agreed it would. – Chris Chinnock
