The application deadline is October 15, 2023.
The Chair of Integrated Systems at TU Munich is working on multi- and many-core processor architectures in application domains like IP network processing, visual computing and automotive electronics. Solutions shall be optimized for various, often divergent objectives, e.g. power efficiency, high performance and fault tolerance. We apply a platform based design approach by reusing already existing design components (IP cores) and complementing the architecture by specific extensions for the intended
optimizations. To demonstrate the effectiveness of our concepts we mainly use prototype implementations based on FPGAs.
In the automotive domain, the steady increase of functional diversity and complexity in today’s and future cars, especially for autonomous driving, require an appropriate upscaling of the vehicular compute infrastructure. A compute hierarchy consisting of near-sensor processing, zonal controllers up to central car servers, interconnected by a Gbit/s in-vehicle network (IVN) with service guarantees are expected to form the future Car-IT backbone. As more and more functions are software-based and with the steady increase in complexity the reliable execution of automotive applications mandatory.
Within the BMBF Project Mannheim-EMDRIVE we are investigating novel methods
and efficient forms of realization for in-vehicle diagnosis that continuously monitors the execution of automotive applications on a distributed vehicular compute platform. The target is to identify potential programming errors or performance bottlenecks undetected during the development process and to track down potential root causes.
Our approach starts with monitoring of the traffic flows on the IVN among sub-functions mapped onto different ECUs/compute nodes within the IVN. In case an anomaly in the expected communication pattern of a traffic flow is detected, tracing of the processing within the source node of the respective flow is activated and the generated traces are analyzed in order to narrow down potential root causes for the anomaly. In our case, we rely on the MCDS tracing infrastructure provided with the Infineon AURIX MCU. This diagnosis process is controlled by a supervisor instance locally in the vehicle, which
is either loosely or tightly coupled with a controller in the cloud that manages the diagnosis infrastructure of a multitude of cars. We do the mentioned investigations in tight cooperation with our industry partner Infineon. Our goal is to demonstrate the effectiveness of the developed concept via FPGA-based prototyping.
We are looking for one full-time scientific staff member for an initial duration of 3 years with the option of prolongation. The position is immediately available. Employment is according to the TV-L wage agreement group E13. Condition for employment is a Master’s degree in electrical or computer engineering. Previous knowledge in several of the following areas is required:
– Heterogeneous multicore processing systems,
– Ethernet communication protocols and associated simulation via Omnet++,
– HDL-based digital circuit and system design,
– Xilinx FPGA prototyping platforms and the associated design flow / tools.
Qualified applicants are invited to send their application documents (CV,
certificates and any other relevant papers) by not later than 15.10.2023 to:
Prof. Dr. Andreas Herkersdorf
Technical University of Munich
Chair of Integrated Systems
Arcisstr. 21
80333 Munich
Email: lis@ei.tum.de
