At the recent TNC24 conference in Rennes (hosted by RENATER, ed.), a dedicated team from NORDUnet used Artificial Intelligence to improve the quality of Wi-Fi and the network in general.
Setting up a network for a conference is like being a referee at a sports event. If nobody talks about your performance, you have done a good job. Still, four engineers from NORDUnet made a noteworthy effort as they set up the network for the TNC24 conference in Rennes, France, early June 2024.
For the first time, the team employed AI in the solution, optimizing the experience for participants.
All conferences, and in particular conferences for highly digitized participants like TNC, see a very high level of cyber activity – live streaming, video-conferencing calls, etc. – with participants logging on and off, often at more than one device, and sometimes moving around.
“The AI based tool helped us monitor all this traffic and suggested modifications, where required. For instance, a participant may begin a Zoom call at one Wi-Fi frequency band but will be moved to one or more other ones during the conversation, avoiding disturbances and latency. Again, the system is like the skilled referee getting the job done discretely,” says Erik Kikkenborg, Chief Collaboration Officer at NORDUnet, heading the team.
Next time: more autonomy
TNC is the primary European event for research and education networks, with not only delegates from across Europe but also strong participation from many other countries around the world. The conference is hosted by local research networks in a different country each year, but always with the NORDUnet network engineering team setting up and managing the conference network.
The successful use of AI for the TNC24 network was nothing more than an appetizer according to network engineer Carsten Pettersson from the team:
“At next year’s TNC, and in fact already at the 2024 NORDUnet Conference (this year to be held in Bergen, Norway, in September, ed.), we plan for significantly more AI implementation.”
An active digital servant
Thanks to technology from the US network company Mist Systems, cofounded by former Cisco Vice President Bob Friday, the team will not only take advice from AI. The switches in the network will be AI based, allowing many optimizations to take place autonomously.
The bot Marvis mini, developed by Mist, acts like a servant to the network administrators.
“Marvis does not just wait for problems to occur. As soon as the conference opens, and people begin their digital activity, Marvis will start monitoring their positions, and which frequency bands they are occupying. The bot analyzes all this data continuously and initiates prophylactic measures when required. And whenever Mist or Marvis sees an outage, a packet capture automatically occurs and will be able to be accessed for forensics,” explains Carsten Pettersson.
Marvis may even act like a virtual participant, for instance simulate a Zoom call to learn the quality of the connection for that.
“All these metrics can be easily accessed by the network administrators, so we will constantly have an overview.”
Setting up in less than 24 hours
Also, the bot will reveal latent problems in the network.
“If, for instance, a physical interface somewhere is not entirely well attached, Marvis will tell us. This saves a lot of time which would otherwise be used for fault detection,” says Carsten Pettersson, adding smilingly:
“Working with Marvis is a bit like having that special colleague on your team. The one that always make things work. A bit annoying but also extremely handy!”
Furthermore, the AI based tool makes setting the network up much easier.
“We would normally start some days in advance, but thanks to the AI assistance we were able to complete the job at TNC24 in less than 24 hours,” reports Carsten Pettersson.
The TNC24 took place in Rennes, France, June 10-14, 2024, hosted by RENATER. The 2024 NORDUnet conference will take place in Bergen, Norway, September 9-13, 2024. NORDUnet delivers networking for both events.
