AV over IP (AVoIP) is the next step in signal distribution. It allows for systems to go beyond the physical limitations of matrix switchers both in input/output capacity, flexibility in that all signals do not have to connect at a central location, and distance limitations for the cabling to that central location.
Audiovisual over IP (AVoIP) has presented quite the conundrum to the audiovisual community with such a disparity of options in flexible signal distribution.
AVoIP is a big step forward for the industry in that it allows projects to get away from large, centrally-distributed solutions, while still fulfilling the need to support immense flexibility in signal routing.
Of course, with so many manufacturers offering an AV over IP solution, there’s now the question as to which one is best for your project? It’s human nature to seek the path of least resistance.
Why make things more difficult than they have to be? The problem with this mentality is that it often leaves us susceptible to honing in on one solution and hesitant to do something different unless we absolutely have no other choice.
AVoIP: A Baseline of Understanding
When AV professionals are working on their AV over IP designs, factors that must be considered are highlighted by every manufacturer:
- Resolution – 4K or 1080p?
- 60Hz refresh rate?
- Chroma – 4:4:4, 4:2:2, 4:2:0?
- Bit depth – 8, 10, even 16?
In the least surprising fact, many of the manufacturers will declare their product “the best” at giving you the optimum video signal. Are they all correct? Does it matter which you pick if they’re all the best? It most definitely does.
AV over IP adds some complexity to the more traditional solutions given the fact that we are now required to play within more stringent bandwidth requirements of the networks.
The greater the digital footprint of an image, the more bandwidth it requires on the network. An uncompressed 4K image at 60Hz with a 4:4:4 chroma subsampling at 8-bit depth requires nearly 18Gbps.
Well, 18Gbps doesn’t seem like a whole lot unto itself in the grand scheme of things, but the thing is, we aren’t worried about sending one video image across a network, we’re potentially thinking of sending dozens, maybe even hundreds across a network. Now, that 18Gbps adds up very quickly.
This is why when we’re looking at the system designs the math is so important. We need to ensure that we’ve accounted for the possibility that each network encoder could be sending a maximum bandwidth signal simultaneously.
Can the complete network backbone we’ve been either given to use by the IT Manager, or the network hardware that we’ll be providing handle that? We must look at the network switch and understand whether it has the backplane capabilities to distribute the number of signals we have connected to it.
We must ensure that the link between the network switches has enough bandwidth to support streams that might be going to different destinations.
If the goal is to send ten 9Gb signals to ten 9Gb destinations, does the network switch support 100Gb of data transmission? If not, how do we need to divide the streams in order to meet the requirements of the system?
Codecs and Compression
So, if each stream could be up to 18Gbps, how is it we’re getting this signal onto a 10Gb or even 1Gb network? That’s where the manufacture’s encoding selection and compression ratios gain relevance.
How each manufacturer chooses to manipulate the image will ultimately dictate the requirements. There are a litany of encoding options that manufactures have gone with – JPEG 2000, HEVC, H.264, VC-2, and even proprietary. (There are so many that a valid comparison of them all requires an article unto itself.)
The other feature that’s mentioned is the amount of compression applied to the image in order to reduce the bandwidth requirements.
The concern here is that the more compression applied to the video, the more likely the image quality, when viewed from the output of the network decoder, will have defects, artifacts, or other distortions that reduce the ultimate quality of the video. However, the less compression that is applied, the more bandwidth the video will require.
Before proceeding to the discussion of some of the more notable AVoIP solutions, take off your “solutions” hat. In this discussion don’t look for “the one.” There is no single winner in this scenario.
Each option has its benefits and each has its drawbacks. Not a single one of them will work for every application on the planet. It’s time to tear down the encampments and look at these for the inherent value in each option and see how and where it fits into your toolbox so that you know when to take them out to make your life easier and your jobs better.
What is it?
“SDVoE is not a manufacturer. We are a technology alliance promoting a standardized approach to signal transport,” says Justin Kennington, president of the SDVoE Alliance.
The upside to this is that any manufacturer that chooses to become a partner in the alliance will be interoperable with every other manufacturer.
This promotes flexibility for those that choose to deploy this solution because they aren’t locked into what Kennington referred to as the “walled garden” of proprietary solutions that only work in one manufacturer’s ecosystem.
Software Defined Video over Ethernet (SDVoE) is based around a common chipset that each manufacturer implements in the products that they choose to develop.
As opposed to similar technologies, like Consumer Electronics Control (CEC), where manufacturers can decide if they want to offer a specific feature or just leave it out, SDVoE locks down these options in order to ensure interoperability between all manufacture partners that choose to implement the solution. To date, there are about 45 manufacturers that have joined the SDVoE alliance, combining to offer 200 interoperable products.
SDVoE is capable of supporting HDMI 2.0 specification, including high-dynamic range (HDR), 10-bit and 12-bit video streams. SDVoE bases their signal calculations on a roughly 12Gbps video stream for a 4K 60Hz 4:4:4 signal.
They reach this by streaming only the active video content, rather than the 33% headroom requirements of HDMI for blanking (a black area around the active picture) and the HDMI encoding used to aid in the transport over low quality cables. SDVoE then uses its own algorithm to compress video streams to 9Gbps. All of this over off-the-shelf network switches.
Why is it different?
SDVoE flips the way to look at how AV over IP and existing network infrastructure work together. Rather than trying to deploy AV streaming devices on a corporate LAN, SDVoE urges that their 9Gbps streams leave open the space for the 1Gb requirements of corporate data networks to coexist on the SDVoE network.
The premise is that because the likelihood of the AV system needing its own network infrastructure is high, why not give the IT Manager an AV network that’s also capable of expanding the 1Gb network infrastructure being used for the company’s data?
AVB/TSN (Time-Sensitive Networking)
What is it?
AVB is a set of technology standards and defined by the IEEE. To ensure bandwidth requirements are not an issue, during configuration of the system, each device is able to set a specific bandwidth reservation.
This means that because each device sets the maximum amount of bandwidth it may need, there will always be enough bandwidth available for the devices connected to the system.
Essentially, once all the devices are connected, the maximum bandwidth requirements are calculated and reserved, eliminating the concern about available headroom in the network bandwidth.
IEEE leaves a lot of options open for manufacturers to consider how they wish to deploy the system. Each is allow to determine factors such as the audio format, screen formats, clocking scheme.
In early AVB systems this meant that where manufacture A implemented AVB and manufacturer B implemented AVB but they didn’t work together because each manufacturer did it in a different way. The AVnu Alliance, though introduced their new offering, Milan, at InfoComm 2018 to solve this problem.
“Milan is a set of rules and agreements that define to use AVB so that devices from different manufacturers are interoperable,” says Henning Kaltheuner, Head of Strategic Business Development and Market Intelligence for d&b and a member of the AVnu Board of Directors.
AVB is known for its rigorous certification process for all products. They have added the Milan certification test to the old AVB certification testing to ensure the interoperability.
Why is it different?
There is no specific chipset for the solution. The AVnu Alliance partners with many chipset manufacturers to develop the hardware. Its bandwidth reservation improves reliability because there is no concern of exceeding the available bandwidth of the devices in the system.
It bears noting that an employee of L-Acoustics has developed a software tool called Hive that’s specifically for integrating multiple manufacturer’s AVB-Milan solutions together to provide for a single software controller. It is currently available on Github.
It must be acknowledged that while Biamp has incorporated video into their product offering, at this point in time the majority of manufacturers including AVB in products are in live sound. When asked about this, Kaltheuner emphasized that the requirements for specific applications are different.
“What we are doing comes from Pro AV [requirements],” (by comparison to broadcast and commercial AV requirements where the option to switch cameras if one goes down, or move to another room if there is a failure), and “what we are doing has to be built in the morning and function by the afternoon.”
What is it?
The latest to jump into the video realm, having dominated the network-distributed audio realm for the last several years.
The technology will be modeled on their existing business structure where Audinate offers a chipset that will allow manufacturers to incorporate a 1Gbps video stream based on either the codec the manufacturer chooses, or using the embedded tools available to the manufacturers within the Dante Product Design Suite.
Since Dante controls the development of the module, that also means that they have retained some control of the features available to the manufacturers. This has proven valuable to them in the past as it aids in standardization and interoperability between manufacturers.
The solution seems to be running along the middle ground of choosing the most popular of the video codecs (JPEG 2000), while also integrating into the Dante controller software.
The new chipset will offer a single video channel with up to eight channels of audio while allowing each signal to be routed independently in the software. Audinate is also touting the HDCP capabilities of the product.
Why is it different?
The upside to this product offering is that Dante has been used widely across the audio realm with great success by many integrators. The downside, though, is that the chipset incorporating video streaming is so new we have no way of knowing what its promise will be.
In speaking with Joshua Rush, Vice President of Marketing and Product at Audinate, he expressed hope that we would start seeing the first manufacturer’s product offering in 2020 at Integrated Systems Europe, but as that is up to the development of the products by the manufacturers (not Audinate) it’s uncertain when that will happen.
It appears like there is potential for great upside here, given the familiarity so many consultants and integrators already have with designing Dante solutions, but until the manufacturers begin to offer products and we will have to wait to see if Dante AV experiences the same kind of broad adoption that the audio solution received.
What is it?
No, HDBaseT is not technically an AVoIP solution in that we are not actually using a network to transport video. Instead, this an AV over a point-to-point structured cabling solution.
This means that network bandwidth requirements are not a concern because we aren’t actually going over a network. This also means, though, that the video transmission also doesn’t have to play by network rules, allowing them to pass larger signals. Per HDBaseT specification 3, they will be able to pass 16Gbps of uncompressed 4K, 60Hz, 4:4:4 video.
Why is it different?
The ultimate promise of HDBaseT is the support of 5-play – video, audio, control, ethernet, and even power over a single cable.
We haven’t quite reached this ability yet because manufacturers have the option to select which features they want to be able to support, but given the new standards for up to 90W of power over PoE, there is a likelihood we may be able to reach this 5-play solution.
Additionally, the new HDBaseT specification three has increased the bandwidth capabilities of HDBaseT’s ethernet component to 1Gbps. This might seem trivial because HDBaseT isn’t, in and of itself, an AVoIP solution. However, think about that feature update from the perspective of what that now adds to AVoIP solutions.
Currently, an HDBaseT matrix has common video inputs and outputs and HDBaseT inputs and outputs. The same will still be true in the future of AV over IP, but what this adds is the ability to now stream video at up to 1Gbps over an IP input or output.
This allows you to eliminate the encoder and decoder that used to be required to bring a streaming signal into or out of the HDBaseT matrix.
Once the new products start coming out, which will likely be in 2020, HDBaseT specification three products should simply the hybrid solution – assuming that we don’t end up in the same situation we have now with multiple, proprietary flavors of HDBaseT.
Why AV over IP Shouldn’t be Your Default
There’s a lot to digest about the AV over IP options available to consultants and integrators, but there’s still one more to cover – why you shouldn’t deploy an AVoIP solution for your projects.
AV over IP offers immense flexibility in signal routing, monitoring of network signals, and opportunity to get more deeply ingrained with the IT Manager. However, just because you can use AVoIP, doesn’t mean that you should use it.
If you’re operating a single room, where signals have no need to ever be viewed beyond that space, there is frequently little need for an AVoIP solution.
This would be an application where HDBaseT would suit your needs just fine because you’re looking for local routing and signal distribution. Where AVoIP has immense added value is when you want to be able to broadcast signals, essentially, anywhere the network is capable of reaching.
Even in those cases, though, that does not mean that a 100% AVoIP solution is the best way to go. You must also consider the hybrid solution where you incorporate local matrices and then provide for streaming options.
It’s infrequent for a room that may have four or five inputs to be able to need to stream them all to remote locations simultaneously. Often, we’re just looking to share the same content that is being viewed locally in the room.
In that case, a single network streaming device being fed from a local matrix would suit the needs of the client just fine.
This is not an absolute, though. There are retrofit projects where a client may simply need an HDMI connection at a table to feed a display on the wall – only no conduit pathway was available and the client wants to avoid surface raceway.
This would be an instance where you could definitely be looking at making the solution completely network based because there will likely be a data port in the floor that can be utilized for the input and one that either exists or can be added at the display.
The fact remains, there is no one single way to look at a project, as see in these three descriptions, and there is not one single way to view an AVoIP solution.
Each offers something unique and valuable depending on the application. Those mentioned are just some of the AV over IP options available. There are also proprietary options, but those come with their own challenges, upsides, and limitations.
That’s why it’s more important than ever to explore the goals of the client and find a solution that matches their goals.
Your options are pretty wide open.