AV-Over-IP: Decisions, Decisions…

By Rob Green and Andrew Starks

You know it’s coming, and you need to be ready for it, but the path to supporting AV-over-IP in your next generation product is still really confusing and choosing a particular protocol implementation feels like a gamble that you just can’t afford to lose!

It’s clear that the benefits of moving to IP-based transport for audio, video, and associated data are becoming better understood by the AV industry. IP networks offer scalability to support any audio and video formats, with multichannel distribution over LAN/WAN and cloud. Network addressable equipment can be the foundation for AV-as-a-Service (AVaaS) and leveraging the maturity and scale of the IT industry to move media around is obviously attractive. However, there is still a lot of uncertainty over ease-of-use, robustness, network management and quality of service, network security and content protection and interoperability between vendors.

As a supplier to a wide variety of equipment manufacturers across the AV industry, we constantly get asked for our opinion about which AV-over-IP standards we think will win out. We saw a similar period of uncertainty and confusion in the broadcast industry, with several competing protocols (not necessarily standards) all vying for the same prize and creating a period of hesitancy which crippled the whole industry. Ultimately, the market figured out that a truly open, interoperable and scalable standard was the way to go, and now SMPTE ST 2110 is the ubiquitous AV-over-IP standard being adopted throughout the industry. In Pro AV, we see a similar pattern emerging. There are several competing AV-over-IP protocols competing for the same space, including (and not limited to) SDVoE, Dante AV, NDI and IPMX, as well as many proprietary IP implementations used in closed systems. All of these have pros and cons, and are all technically excellent, so which one should you consider?

Openness, and the reality of interoperability

The Pro AV industry is different to the broadcast industry in that there is often a tendency to tolerate and use enclosed systems from one supplier because “it will just work.” However, like the broadcast industry, system integrators and equipment manufacturers are reluctant to be held hostage to one supplier, even if it works well. Being able to select best-suited products for a specific use case on part of the network and seamlessly integrate that with another manufacturer’s products on a different part of the network is really what truly open IP networking is all about. The same is true about the AV-over-IP protocol selection. Take SDVoE for instance, which claims to be interoperable but requires a specific Semtech ASIC device to be used for every IP interface. If you own both ends of the link and communicate with yourself, can you really claim to be interoperable? Using a vendor-specific device also creates risk – what if that vendor decides to increase pricing (which they can because there’s no competing supplier for that implementation), or worse, stop supplying that device to the market because the return on investment declines? What is needed for a healthy community to prosper is an open standard, accessible to all that can be implemented by anyone, on any capable device.

Scalability and adaptability

ASIC implementations suffer from the inability to adapt, which you really (really!) need in a chaotic environment such as AV-over-IP. If an ASIC supports a fixed rate such as 10Gb Ethernet, you can’t scale or adapt to different use cases. You need the fastest link that makes economic sense. This could well be 10G if you’re building a greenfield install, but most of the market is still at 1G and would prefer a technology which offers high quality, low latency compression to make use of the installed network, rather than rip everything out and start again. If you’re going to start again, why limit yourself to 10G – why not go 25G, 50G or even 100G and make it future proof? Bear in mind that for multiple 10G links to handle multiple 4K streams, you’re likely to need higher-speed backbone network anyway. 8K displays and TVs are already on the market, so if you eventually want to move to 8K and are limiting bandwidth to a specific rate, it may well require multiple devices, different compression, or another new “standard.”

IPMX – the only open and scalable AV-over-IP standard

IPMX is based on the proven SMPTE ST 2110 standard being deployed throughout broadcast, but it addresses some additional needs of the Pro AV industry, including HDCP copy protection, network discovery and registration, I/O management, and the enhanced audio channel mapping required for specialized systems, including multichannel surround sound. IPMX is scalable and supports any rate with compressed or uncompressed streams and is video format and bitrate agnostic.

Perhaps the most important feature of the standards and specifications that comprise IPMX is that anyone can download the complete set of standards (document fees and codec licensing may apply) and then build anything they like with them. That is what a truly open standard means. Every manufacturer making products that conform to the standard is on the same footing. Everyone in the community is striving to ensure that all equipment interoperates and that there are test suites and testing equipment available from multiple vendors to prove it works. IPMX is implementable on any capable device of your choosing, so no being held hostage. Obviously, we’d be delighted if that was a Xilinx programmable SoC, and it’s highly likely there will be an FPGA or SoC in your product anyway, with Macnica Technology IPMX solutions available to help you, but this is more about open standardization to benefit the industry.

Fig. 1: Macnica IPMX Module and Development Kit.

ST 2110 is proven and deployed, and as SMPTE, AMWA, VSF and the AIMS Alliance add further AV-specific features to IPMX, it can leverage the software platforms and programmable SoC devices that are likely to be already in your system for updates. It’s still the standard to choose if you want least risk and a future-proof AV-over-IP implementation. It’s true that one-size-fits all is often unrealistic, and there will be use cases where NDI, Dante AV, SDVoE, etc. will make perfect sense, but once IPMX is fully rolled out, it seems inevitable that the market will gravitate towards it. Either way, the market always decides!

Rob Green is a senior manager for Pro AV, Broadcast & Consumer at Xilinx.

Andrew Starks is the director of product management at Macnica Technology and an AIMS board member.