CeCaS

Highly automated vehicles must process huge amounts of sensor data in real time. CeCaS (Central Car Server – Supercomputing for Automotive) developed a supercomputing platform for the vehicle for this: instead of many separate control units, a central computer controls the vehicle's software functions.

• Moving from a domain-based to a zonal, redundant, high-availability architecture
• New heterogeneous computer structures with high performance at low energy use, built on new automotive-grade high-performance processors in FinFET technology
• Application-specific hardware accelerators, adaptive software and neuromorphic approaches, designed for the highest safety level, ASIL D
• Adapted packaging, interconnect and cooling for robust use in the vehicle

Our contribution

The CeCaS team at the Institute of Computer Engineering designed, among other things, a memory protection unit that prevents unauthorised access and a trustworthy AI accelerator. In addition, a virtual simulation platform lets various parts of the system be simulated and analysed for security. More in the university's press release (in German). Results include the SCA-WAL framework for assessing power side channels (IEEE Embedded Systems Letters, 2025) and an architecture for a hardware root of trust (Journal of Cryptographic Engineering, 2025).

Partners and funding

28 partners coordinated by Infineon Technologies AG, among them Bosch, Continental, ZF, CARIAD, AVL, HELLA, KIT, FZI, TU Munich and several Fraunhofer institutes. More on the project website.

Funded by the German Federal Ministry of Research, Technology and Space (then BMBF) in the MANNHEIM programme of the Zukunftsfonds Automobilindustrie, grant 16ME0800K.