In search of ever more efficient use of human and technology resource, many broadcasters are creating distributed (multi-site) teams & multi-purpose systems & facilities and utilizing remote production workflows. IP is an enabling technology for this.

Many are also incorporating private data center based, software first systems to bring greater flexibility and scalability, and here too, IP is an enabling technology. The software first model also extends to use of SaaS and FaaS software systems, hosted on either private data center systems or public cloud systems - to bring readily scalable licensing models to software systems.

As central as IP may be to these new infrastructures, the reality for many is that legacy baseband SDI equipment still has a lot of life left in it and remains a financial asset that needs to maximize its ROI. System operators are familiar with this time-tested technology and its workflow, and software-based systems can be difficult to deploy in a timely fashion. Broadcast engineers are still trying to finesse how to make all of the devices on an IP network easily accessible and communicate with each other, and therefore simple for operators to access and use.

SDI is a mature, standardized technology that is cost effective, provides frame-accurate switching, fixed low latency, and synchronized video and audio - among other capabilities that are critical to making, moving, and monitoring video and audio signals in live, real-time applications.

As time passes the adoption of IP is becoming easier as familiarity with IP improves, the cost of off-the-shelf commodity IP switches and other IP networking technology falls, and 40-Gig-E and 100-Gig-E infrastructures become more prevalent. This brings greater bandwidth, which is necessary to carry data-intensive signals like UHD. All-IP environments certainly do exist and are working incredibly well, but most installations that have gone IP still have some SDI involved somewhere. Indeed, there are few situations where there is an absolute “no SDI” policy because there are few business cases that can afford it.

IP clearly has enough compelling advantages to make its adoption worthwhile but are we really on a complete migration path from SDI to IP?

‘All-IP’ Networks Are Challenging

“In general, IP networks are more challenging to set up than SDI because there are multiple layers of networking to deal with,” said Kurt Heitmann, CEO, CP Communications, a mobile production truck builder with locations across the U.S. “It requires someone that can manage the network router, the switch, the gateways and the ports inside of the networks. Then you need to create VLANs for video, for Dante audio, and for communications, and all your assignments to those networks have to be correct. If you have two conflicting IP addresses or an IP loop, the whole system goes down. SDI networks don’t require the same level of configuration. It’s going to be 720p, 1080i, or 1080p. That’s all there is to it.”

“In 12G systems you have one wire for everything from SD to HDR, so the production infrastructure is so easy to install and maintain,” said Renata Mancini, CTO, ARET Audio & Video Engineering, in Italy. “IP networks are a bit tricky to get right, especially when combining it with an existing SDI infrastructure.”

An Incremental ‘Middle Ground’

As typifies the pragmatic approach of broadcast engineers, the evolution of broadcast systems is always slow, cautious and practical. Looking back at history may feel like systems rapidly move wholesale from one technology to another, but things rarely do actually move quickly or in great technological leaps. Change is modular – new technologies are adopted in incremental steps, usually only when they can slot in reliably to existing infrastructures.

So it is with the move from SDI to IP. It is entirely possible that over time SDI will disappear, but over the last few years, it seems that for many the focus has been on finding ways to embrace the advantages of IP, whilst retaining all of the advantages of SDI.

The inevitability of that incremental approach to adoption necessitated the creation of ‘glue’ devices to bridge and connect the two domains. Logically, this brought about the use of IP as the routing and transport backbone, with islands of SDI within different parts of the system. It turns out the approach is very effective. The result is scalable hybrid systems that feature both – and increasingly also feature software-defined, data center based systems.

Adding Devices To the Network

Once a networked SDI-IP environment has been established, how do you manage which devices are available to which control rooms and production teams?

Allen Harris, Lead IP & Bonded Cellular Specialist, CP Communications, said that it really depends on the situation. The key is to fully understand the network you are configuring and how your production team likes to work. Then begin to add devices and design work groups of equipment resources for specific use cases.

“If the control rooms are in different places, we assign devices to output devices,” he said. “That provides access to the signals on their playout servers. Depending on what technologies we use, we configure it all in the cloud to set up streams to go direct to clients, consumers or the location that needs to receive them. There is a lot of variety based on which technologies we use and where the content ultimately needs to go.”

NMOS Helps Find Devices On The Network

When a new device is added, technologies like NMOS and Ember+ help to establish an IP address on the network and, accordingly, to recognize devices on the network.

“NMOS is the protocol of choice,” said Rainer Kampe, CTO, Broadcast Solutions, German systems integrator and builder of OB Vans. “Through the use of NMOS our Human Interface control system can automatically detect the device, assign multicast addresses to the stream senders and add the nodes, senders and receivers to the system. The user can then activate them.”

“It’s never just plug-and-play,” said Harris. “We typically have to introduce the device to the network. Some video technologies we use make that process easier, requiring no intervention from a third party. We can do everything ourselves, but we have to point to a server and manage it. Once the device is in the system, you turn it on, and it’s up and running. Another system we use allows us to just add it to our portal and manage it, which is straightforward.”

For Dante-compatible audio networks, audio production devices that sit on the edge of the network are configured to be compliant with the Dynamic Host Configuration Protocol (DHCP), which makes adding an audio device fairly simple.

“When an audio device is introduced to the network, we provide a place to plug it in so that it is on the correct network,” Harris added. “It will get an IP address, and will be recognized by the system and made immediately available. That said, operators still have to do some drag and drop configuration to get the audio sources or destinations to and from that device.”

Handling UHD-HD up/down/cross conversion is also a factor to address. Typically, some type of hardware/software converter is used. Harris said that if they are trying to cross-convert from HD for distribution, they’ll use a hardware converter. If they’re just looking to monitor the conversions, they will often use a smaller, more cost-effective converter to provide a way to look at that image on a screen.

Compression Is Key To Signal Distribution

Many workflows, especially when remote locations are involved or a studio has lots of content in its native form, necessitate some form of compression. In most cases it’s H.265 or JPEG XS.

“If it is a situation where we are trying to deliver content to someone who wants to stream to the web, we will use H.264,” said Harris. “This is because H.265 is not a format that most re-streaming environments like YouTube and Facebook want to see. For some sports like auto and motorcycle racing, we will use H.264 because there are certain aspects of H.265 encoding that don’t play as well in those very fast-paced environments.”

He added that occasionally they will run both compression formats because of the way that different things react in different situations. He said they look closely at the content and how the end users are using it “as opposed to forcing them into a technology. We will evaluate how things perform and make adjustments to provide the best results with the best encoding.”

Broadcast Solution’s Kampe said that the determining factor for which compression to use comes down to the available WAN bandwidth.

“If sufficient bandwidth is available, JPEG XS is the preferred choice,” he said. “If not, various flavors of H.264 are used. Most importantly, a framework should be utilized that allows for ultra-low latency encoding/decoding and synchronized ‘multi-camera’ sessions.”

Software Control Layer

In many larger facilities, there’s a constant need to configure and reconfigure signal paths as required for specific projects. This is where a software Broadcast Controller layer that supports multi-vendor deployments is essential. It also helps a production facility smooth its transition to full IP operations in the future. And it can automate the often-complex network IT setup process, saving time and human resources.

There are many choices offered by the industry, but they all offer some type of Web-based system integration toolbox that allows the user to easily change interfaces of individual devices or the direction of signal paths. This allows the infrastructure to adapt and grow as needs and requirements change over time, whilst enabling the user to manipulate the network from a single interface screen. They also seamlessly integrate with the majority of widely used baseband and IP broadcast equipment. And these software control systems provide system-wide snapshots for recalling and scheduling recurring or most desired workflow configurations.

“This is the most challenging part,” said Broadcast Solutions’ Kampe. “Each broadcast control system has its own approach. You can spend hours and even days defining the workflows and mapping them in a broadcast controller. It doesn’t really matter whether it’s an SDI, IP or hybrid environment.”

“In an ideal world, a standardized control procedure would be the goal,” he added. “However, this would require the harmonization of the ideas and protocols of all manufacturers. We are a long way from achieving this. A control system can integrate numerous APIs and protocols, but beyond a certain point, the effort required becomes too great for everyone involved. Therefore, in addition to the control system, you will still need to work with the manufacturer’s GUIs. Our credo is: ‘We support and automate everything that works with NMOS!’”

Designing Flexibility Into The Network

Of course, one of the main purposes of deploying IP networks is to easily reconfigure the infrastructure to manage resources and support different production projects. Ensuring sufficient capacity needs to be planned in advance: If a network is all IP, expect to need a minimum backbone of two redundant 10Gb networks.

“We use a multi-gig network approach,” said CP Communications’ Heitmann. “Our networks typically have a 10Gb backbone, meaning that everything is connected. All the switches are connected by 10Gb links. And each device connecting to that uses 1Gb connections.  If there are connections moving between different switches, there is no contention for bandwidth. It is built to take multiple high-bandwidth connections simultaneously, even between different parts of the network within the different switches.”

From a scalability point of view, he said that if they are looking for a solution that requires a significant number of endpoints, they reconfigure the network tool kit that they’re using, from all hardware-based solutions to data center based software solutions, to enable higher amounts of connectivity. “That also gives us the ability to connect to a wider area with less chance of the network connectivity being negatively affected by multiple connections.”

Broadcast Solutions’ Kampe said they have a slightly different approach to system scalability.

“Designing a system as a monolithic block is a dead end,” he said. “Any extension would require the entire system to be modified. Therefore, we adopt a clustered approach to scaling systems. We create individual, independent ‘islands’ around a centralized network built on standard IT design principles to ensure scalability. These islands, surrounding the central network, can be either IP-based or SDI. Typical examples of these islands include control rooms, studios, playout facilities, etc., or even cloud-based systems that can be spun up and connected.

“This method resembles a spine-leaf network and adheres to the same design principles,” he said. “The ‘spine’ is the IP network, while the ‘leaves’ are the different production facilities.”

Building Hybrid OB Vehicles

OB vehicles can benefit significantly from reduced cabling, higher density routing and flexibility in the IP world, but it seems that for the majority of OB vehicles, SDI is nearly always still involved. While there are IP versions of most products, we are told that some of the freelance community still needs to learn how to use the new products or embrace the new workflows. Perhaps this is unsurprising given the level of muscle memory required for the fast pace of live sports production. Progress is ongoing but adoption challenges remain.

Deploying a hybrid infrastructure on board an OB Vehicle makes just as much sense as it does in a fixed installation. The differences between a fixed studio or mobile deployment remain minimal for video, according to CP Communications’ Heitmann.

“Most trucks and studios work with SDI baseband signals,” he said.  “We have not seen much complete conversion to ST 2110 in either environment. Audio is a very different story, as mostly everything runs through Dante, especially on board trucks. That also applies to intercom. Studios vary a little more, though newer production studios are all over-IP now when it comes to audio.”

Conclusion

Broadcast equipment manufacturers have greatly expanded the availability of products that support SDI and IP simultaneously and help optimize the bridges between the two. Broadcast engineers are becoming more experienced with IP and how it can interact with an SDI workflow. A practical balance between the two approaches has been found with the creation of hybrid SDI-IP infrastructure.

There are many reasons for maintaining a hybrid SDI-IP environment. For most broadcasters, growth plans are always being evaluated. Expansion of channel count, broadening system or network capacity, adding new services, and upgrading to new resolutions and formats is inevitable, but the pace of change and available budgets are not fixed.

IP fully and easily supports all that and more, and leverage’s the many benefits of widely available COTS hardware. At the same time, the pace of change for most broadcast facilities is historically slow and incremental and SDI is still a practical, familiar and extremely reliable solution, co-existing with IP.

SDI may also remain the best choice, for now at least, for certain types of equipment and technologies, or for organizations of certain sizes and budget levels.

The use of SMPTE ST 2110 makes it easier to manipulate separate video, audio and ancillary data streams, but no one is writing a death warrant for SDI any time soon. There are still situations where SDI outperforms IP. SDI is still lower in cost, easy to trouble shoot, ties into existing infrastructure, migrates easily to UHD and can carry uncompressed signals without latency.

With a hybrid approach, system designers are intelligently taking the best of what both SDI and IP have to offer and creating robust, efficient, scalable and highly cost effective infrastructure. This hybrid philosophy is already encompassing incorporation of cloud-based and new microservices software based systems that are set to further improve the versatility and scalability of broadcast production systems.