SVGE Analysis: IP live production power grab and the aims of AIMS
Despite vocal efforts to avoid launching the new era of IP based television as a minefield of proprietary and inoperable systems, broadcast manufacturers seem to be doing just that. There are already several transport protocols, with overlapping and competitive components, which are in danger of creating customer confusion, or worse, leading them up a blind alley.
One clear division is whether to work with video over IP in live environments as compressed or uncompressed media. And, if compressed, then which codec is best?
Collectively the industry would seem to want one ring to rule them all but since this would give the winner a considerable amount of power, commercially that’s worth fighting for. This ambition puts customers and the industry at risk of being entrenched once again in islands of production. This article explains how.
The contenders
The contenders include the TICO Alliance, ASPEN, AIMS and a proposal from Sony. They variously support, rework or sidestep standards like SMPTE 2022-6 and the work of bodies like Video Services Forum (VSF), EBU and Advanced Media Workflow Association (AMWA).
Matters are further complicated by the spin of marketers who claim that their approach is fully open to third parties and standards based, claims which can be disingenuous. For example, all contenders use the muddy waters of RDD’s (Registered Disclosure Documents) where technical recommendations have yet to be ratified by a standards body. Vendors can use RDDs to justify claims of a standards-based approach while not being fully transparent that these are in principle only.
The business core of the issue depends on who you talk to but manufacturers each have a vested interest in selling more of their kit if certain video over IP routes are chosen, whether those routes are proprietary or open.
The aim of AIMS
The newest kid on the block is the Alliance for IP Media Solutions (AIMS) with Grass Valley, Nevion, Imagine Communications, SAM and Lawo as founder members. It’s a lobbying group for ‘support of open and publicly available protocols for moving media over IP’. It supports the work which AMWA began last September as the Networked Media Incubator (NMI) and which builds on the work of the EBU/VSF/SMPTE Joint Task Force for Networked Media (JT-NM). AIMS attaches particular importance to VSF technical recommendation TR-03 and, by association, TR-04.
“AIMS collectively — and SAM individually — see a danger in the adoption of proprietary formats,” says Tim Felstead, Head of Product Marketing, SAM. “Our philosophy is that we should be adopting open protocols that are as much as humanly possible royalty free.”
“There’s always a danger [of fragmentation] and it’s the reason we formed AIMS,” explains Steve Reynolds, CTO, Imagine Communications. “Instead of each member setting their own pattern to manage the transition to IP we made this joint decision to adopt a set of standards which 70+ members of the VSF have been working on for several years.”
Reynolds describes AIMS as a trade organisation. “We’re not a standards development body. We are not building new or proprietary protocols but committed to open and interoperable tools.”
TR-03 versus SMPTE 2022-6
The TR-03 recommendation promoted by AIMS and devised by the VSF is similar to SMPTE 2022-6 with the crucial difference that while 2022-6 multicasts audio, video and vertical blanking essence in a single uncompressed stream, TR-03 passes individual streams through a network, to be re-composed into different combinations as needed for production purposes. An example might be connecting different audio tracks with a video.
“2022-6 is a point to point protocol and doesn’t change anything to benefit of customers,” argues Felstead. “If you want video to go to one place and audio to go to another then you have to take 2022-6 through a process to de-embed the audio and send it to another location. The characteristic of TR-03 is that you start with three different transport streams (audio, video and ancillary data) and those three are then handled separately.”
TR-03 carries audio using AES67 and has its network clock distributed using IEEE 1588 Precision Time Protocol (PTP). Streams and synchronized groups of streams are described with IETF Session Description Protocol.
The work by the EBU in using 2022-6 to create a live IP studio at VRT in Belgium was certainly groundbreaking in creating an interoperable network but is seen as just a first step for the industry into IP and not necessarily the right answer going forward.
“2022-6 brings us no further forward than SDI,” says Felstead. “Will we go back to the transition between linear and nonlinear where the industry substituted tape machines for servers and then just carried on in a linear workflow? If we take SDI into the IP domain we are guilty of making the same mistake, whereas the more adventurous way forward is to split the components and try to devise a better and more efficient workflow.”
It would be nice, he adds, to “avoid the painful birth that we had with networked audio standards [where ten different audio formats delayed widespread implementation by a decade and finally culminated in AES 67] to video over IP.”
The ASPEN approach
Adaptive Sample Picture Encapsulation (ASPEN) creates a framework to carry uncompressed video over the transport stream and makes use of components of MPEG-2 to do so (including for synchronisation and the transport of audio). This is currently documented via RDD 37 and as SMPTE ST 302 (for audio) and 2038 (for ancillary data).
Led by Evertz, ASPEN is also being implemented by ChyronHego, Vizrt, Tektronix and PacketStorm. Cinegy, I-Movix, FOR-A, Hitachi and Matrox are also among 30 backers of the solution. Existing Evertz customers Dome Productions, Game Creek, NEP Group, Time Warner Cable Sports and Discovery Communications have deployed ASPEN. Interestingly, Sony is also a supporter with a key joint customer deployment at NBC Sports.
The Sony Networked Media Interface
Cisco, Evertz, Imagine, Matrox, Rohde & Schwarz DVS and Vizrt are among 42 backers for Sony’s live IP production protocol branded Networked Media Interface (confusingly with the same acronym – NMI – as the AMWA working group of which Sony is also a member).
Sony NMI is an adaptation of SMPTE ST 2022-6 and ST 2022-7 designed for the live environment. It supports the SMPTE standards for transport of uncompressed HD-SDI over IP as well as SMPTE 2059 PTP for timing. It diverges from AIMS in supporting the transport of media wrapped into a single multicast stream; it does not support the AES67 standard and it deploys its own Low Latency Video Codec for 4K 60p transmission over 10 Gig Ethernet. This technology is before SMPTE as a RDD34. Sony continues to participate in the development of IP technologies at the VSF, AMWA and EBU.
“To be honest I do not think there is a risk of industry fragmentation,” says Nicolas Moreau, Product Marketing Manager IP Live Production & Workflows at Sony. “If you look back to when we first introduced NMI it could have been seen as proprietary solution but Sony has quickly opened up RDDs to SMPTE and other groups. We are committed to an interoperable approach.
“With regard to AIMS — we don’t see them producing standards and that is our goal,” he says. “TR-03 is not a standard yet, it is only a recommendation and we only support standards [though Moreau neglects here mention of Sony’s own draft RDD34]. “To support AES67 you need product that meets that standard and as yet there is none.”
Moreau says his main concern is the undue focus on SDI to IP mapping and transport “that the customer tends to forget all about the other vital functions for doing production over IP,” says Moreau. “These include dynamic routing, device discovery and integrating software defined networks. Much of this is also forgotten in the proof of concepts we see on the market. NMI is offering a path to solve that.”
Considered proprietary by some, Sony says it is far from shutting out industry developments. For example, Moreau says that audio is one of the directions Sony is working on. “No-one is saying that any IP solution will be rock-solid for the next decade. Standards evolve and when the industry has something to look at we will take care of it. We are delivering FPGA IP cores today which could be updated if a new standard emerges.”
The TICO Alliance
Like Sony, IntoPix has devised a codec for compressing 4K over 3G-SDI and over 10Gig IP-based network. This is supported by The TICO Alliance, a consortia of organisations including Grass Valley, EVS, Imagine, Nevion, Ross Video and Tektronix which says it is working in “an open and collaborative way with industry organisations, including SMPTE, VSF, and JT-NM, to guarantee an interoperable adoption.”
For some members of AIMS, though, the approach of Evertz, TICO and Sony cannot be supported. “There’s no doubt these are good technologies but our biggest bugbear is that they are proprietary and licensed,” says Felstead. “For that reason we believe it’s not sensible for customers to adopt it.”
Compressed solutions count
While SMPTE 2022-6, TR-03 and ASPEN offer uncompressed solutions, SAM’s Felstead argues that simple economics make compression necessary. “Moore’s law would suggest that bandwidth is growing ad infinitum but it still costs to develop routers and infrastructure that functions together at low latency so in economic terms, uncompressed is not automatically better,” he contends. “You need some form of compression in order to reduce bandwidth consumption.”
Even though SAM is a backer of Sony’s NMI, it want to encourage adoption of the BBC created, VSF supported, royalty free VC-2 codec for compression within a TR-03 industry agreed interoperability framework. Like TICO, this is wavelet-based.
“For the time being TR-03 is about uncompressed and as yet is not dealing with compressed standards,” says Felstead. “That should be dealt with in line with the same philosophy and SAM are already seeing that in our products today with VC-2 compression and TR-03 streams of switchers, routers, modular interfaces, servers, playout, and others.”
The AIMS roadmap does indeed accept use of compressed as well as uncompressed bitstreams.
“If end-users want to use TICO or Sony compression, or for that matter AVC, J2K or VC2, in an AIMS framework that is valid,” says Imagine’s Reynolds. “We don’t see friction with what Sony and TICO are doing.”
However, according to Reynolds, the ASPEN approach is incompatible. “Evertz came up with a clever solution to doing IP encapsulation in the transport stream,” he says. “They launched some projects for ESPN and NBC and needed to come up with a framework to solve the move to IP and decided to do that with a suite of [MPEG-based] technology that was well understood. Nonetheless, this has compromises that are not aligned with the longer term vision of the industry. It’s evident that ASPEN was a short term solution to a short term problem. That’s not to say Evertz won’t move into TR-03 but we didn’t see the need to take that interim step.”
The future of AIMS
Cisco, EVS and the Telos Alliance have recently joined AIMS but Felstead says there’s no arms race. “I don’t thinking joining AIMS is necessarily the benchmark of success. Adhering to its philosophy is the important thing.”
Reynolds reiterates that Imagine wants to have the same level of guaranteed interoperability with IP as with SDI and that this involves many different manufacturer’s technology being connected.
“This is possible as long as there’s an adherence to the formula in the AIMS roadmap,” he says. “We believe we can build a momentum broad enough to avoid the balkanisation that bedevilled the move into file and to create an interoperable formula from day one. Anything that diverges from the work done in the VSF would, I believe, be detrimental.”
While there are clear overlaps and lines of compatibility between these groups, the industry will surely be looking for greater coherence from the industry heading into NAB.