Artificial Intelligence has the potential to process high-dimensional data very efficiently. We investigate how concepts of Deep Neural Networks or Convolusional Neural Netowrks can be applied to embedded systems.

To cope with the rising performance requirements of future embedded systems we are developing new energy efficient hardware- and software solutions. We specifically focus on devices that incorporate reconfigurable and adaptive hardware.

IoT systems often consist of a great number of networked embedded devices combined with a number of software services in the Cloud and Edge. Management and control of such highly distributed systems is a interesting challenge for IoT systems.

IT security is a special challenge for embedded system, due to their limited amount of resources. A potential solution for IoT applications is Physically Unclonable Functions (PUF) on FPGAs.

Future network architectures are expected to meet diverse service requirements. One way to overcome these challenges is to adopt Network Function Virtualizatio, Software-Defined Networking and Multi-access Edge Computing.