The network as system adhesive


9 June 2014

In today’s world, a single computer system often isn’t enough anymore to get a job done; users often have to deal with several systems – which are rarely integrated with each other. Technolution has developed a solution to digitize and combine diverging computer systems of this kind with each other. The challenge was to come up with a solution that would be able to deal with all the different situations that may occur.

Article from Bits&Chips #6, June 2014 – also available in PDF (Dutch only)

Users are increasingly dealing with systems of systems: several computer systems with different specific functions that are linked together to provide a total solution. Modern operating theaters are a good example. During an operation, the surgeon uses imaging systems such as X-ray imaging, systems to monitor vital functions of the body and microscopes. This trend makes it increasingly necessary to integrate user interfaces with each other, for instance by flexibly displaying information from the different systems on a single large screen.

But the principle of integration is rarely given high priority when subsystems are being designed. This means that the processing of image and video streams can be problematic, and the various user interfaces are often difficult to integrate. The result is: workplaces with many different monitors and loose keyboards, and users who have to roll from keyboard to keyboard on their office chairs in order to manage the system of systems.

What makes it even more complex is that information nowadays really has to be available not only on one central location, but anywhere and everywhere – and this includes video streams. A hospital, for instance, will want to be able to show the images of an operation live in an auditorium. Therefore it has to be possible to distribute video streams flexibly to many different locations. The requirements in respect of image quality and maximum latency differ from case to case, but the call for flexibility and distribution of video streams can be heard in many diverse fields of application.

At the same time there is a growing supply of video streams that come into the system. Take a look in a traffic control center, and you will see a huge quantity of images from roadside cameras. These images have to be collected and the relevant video streams subsequently displayed on the monitors in the traffic control center. Dozens of images come together in a single workplace.

If we could begin with combining systems from scratch, we would be using standards to guide the linking of video and interfaces. But in practice we’re dealing with existing, closed systems without APIs to build combined solutions.

In order to solve this problem, we developed SigmaXG at Technolution. Put simply, SigmaXG is an advanced 10 GB Ethernet infrastructure. Video, audio and other data are dispatched together through the network. Systems can be connected to this network using SigmaXG boxes that support all current video connections such as VGA, DVI, Displayport and HDMI. Because it digitizes USB, mouse, keyboard and memory stick can also be connected to the network through the box.

In SigmaXG, the network is the basis for exchanging video, USB, audio and other data. This network approach makes it simple to add intelligence by linking up with other systems; in this way you could add image storing or combining of video from various sources, for instance, or a system that understands the workflow and can display information at the right moment and in the right place.

In addition, SigmaXG has a video mixer, which makes it possible to display the video output – including the user interfaces – of several devices together on a single screen. The mixer also ensures that all mouse and keyboard input on this combined screen is sent to the right system, so that the user can operate all connected devices from a single workplace.

Less demanding

We already had experience with many partial aspects, such as introducing video on an FPGA, building networks and processing video, but our solution raised all kinds of technical challenges. The most important one was that it is necessary to keep the right balance when you are combining systems into a larger whole: this has to be generic enough to be applied to different domains, but it also has to be built with enough knowledge of these application fields to be able to offer added value. A solution for the traffic market, for instance, is not necessarily suitable for application in an X-ray system for operating theaters.

In operating theaters, compression (which involves losing information), is entirely unacceptable. Moreover, video has to be sent across the network with minimal latency; while the source is in the process of dispatching the image lines, these are already being displayed on a monitor a few milliseconds later. In order to realize this, the clocks of the video source and the monitor have to be synchronized precisely. This has to happen over the network, without a physical link between the clocks. This sharp timing can only be realized with an electronics and VHDL solution on an FPGA. We also developed a special method to code the information on the network. This means that SigmaXG can send HD video from an X-ray system across a 1 GB connection at sixty frames a second without compression.

The requirements are less demanding in a traffic control center, but here there is a much larger number of video channels that have to be processed, while the system also needs to have compression functionality. We were able to develop a software solution for this field of application. This usually makes it quicker to realize functionality, so that we could use an Agile development process in which the client is involved in devising the solution.

Of course it’s a permanent challenge to realize performance and functionality with a cost-effective architecture. The choice of key components plays a crucial role in this. Together with suppliers, we were able to choose components that are available in different types with different prices and functionalities. Our solution, moreover, is modular: the network can be connected with the interfaces that are needed for specific applications. This means it’s not a one-size-fits-all product half the functionality of which is never used, but a product that offers application-specific solutions on the basis of a broad basic platform, into which, for instance, knowledge of video compression algorithms has been integrated.


Jacco Wesselius and Adriaan Schipper. 

Editor: Pieter Edelman