Self-troubleshooting network system thanks to artificial intelligence

BT telecoms company (UK) is testing the network of human-like reaction functions such as self-tuning, recovering from incidents, flexibly following command changes . so that the network has higher stability .

At the Artificial Life XI conference in Winchester, experts in biology, robotics, computer science . discuss ways to bring ideas from life into hardware and software from the beginning.

"If you look at the biological world, people see a lot of changes, complexities and adaptations," said biologist Paul Marrow in BT's Broadband Application Research Center. "This world is the inspiration for us to make such delicate and flexible products and services".

BT hopes the future network will be "more autonomous".Photo: BBC.

For the telecommunications industry, explicitly dividing tasks and allocating them on different components of the network is called "packet encapsulation" and is like the function of living cells in the muscle. can. BT has teamed up with Telecom Italia and several universities to implement the Cascadas project to optimize the online division of tasks by giving them self-awareness and "autonomy" capabilities. The tasks will be easily compatible with the various programs that can call them and increase performance.

"This self-organizing model will take very simple rules to separate units and they work together to do a bigger task, just like in the ants," said Fabrice Saffre, expert of BT, explained. "For example, when a server is requested by multiple users for a similar task, it sends a message to the nearest" neighbors "in the network to find out where the application is needed. "neighbor" server will notify the nearby computer and so on until it finds out to do the job. "

Thus, the server will be able to independently load the load for it and only contact the servers directly linked to it, not "issuing commands" for the entire system, causing time and wasted talent. original.