Will enable intelligent machines to interact wirelessly

The current network for wireless communication, 5G, is primarily optimized for connecting people. In the networks of the future, autonomous vehicles, robots and other AI-based systems will also need to communicate. Wallenberg Academy Fellow Zheng Chen is developing methods to make this possible without overloading the network.

Many researchers are working towards enabling intelligent machines and robots to take over tasks now done by humans. In some parts of the world, for example, self-driving cars are already on the roads. However, for intelligent machines to be used on a large scale, they must be able to exchange information wirelessly. They need to be able to make shared decisions and solve tasks together in real time.

Current wireless networks and models for artificial intelligence are not constructed for this. Today, when data are analyzed using artificial intelligence, all the data are collected on a single server where the analysis is conducted. This method works for moderate data volumes, but another solution is necessary if thousands of cars are to drive around a city without crashing or causing traffic jams.

Dr Zheng Chen at Linköping University is developing new principles for communication and machine learning, which will decentralize artificial intelligence. The aim is for AI-based machines to be able to analyze and process data locally, in their own device, while simultaneously exchanging information with other devices. Learning will take place between devices, so that machines can solve tasks together without being dependent on a central server. 

Photo: Patrik Lundin