Electronic systems and VLSI design

In this theme, we facilitate a performance conscience methodology to realize digital and mixed electronic systems (e.g. SoC's, multi-core, dataflow DSP) starting from high level algorithm descriptions. Further, we use this methodology to realize typical complex computing blocks, for example security and safety hardware, multi-core processors, network interconnect, neuromorphic compute blocks, high-speed communication interfaces or similar complex compute blocks.

The design problem to be solved is how to design, connect and implement large IP blocks, in the 'best' possible way, i.e. in terms of speed, bandwidth, power consumption and data reliability. Topics to be covered among others low power optimization and reduction techniques, SoC design methodology, modeling, specification and implementation, communication architecture and protocols.

We work together with a large variety of partners: Medical, Telecom, Security, and technology companies. MSc projects are available in these domains.

Several languages have been developed over the years to support a designer in his design tasks. Most prominent is VHDL and more recently SystemC has emerged. Those languages provide a syntax; how to use such a language in an efficient way is an important research topic. The designer has to make choices between different architectural solutions, different partitions, etc. Typically he is concerned about speed, power, and area. How to find a solution which fulfills his requirements?

Projects under this theme

Automotive Intelligence for Connected Shared Mobility

Architectures for embedded intelligence and functional virtualization for connected and shared mobility using trustworthy AI

Distributed Artificial Intelligent Systems

Running existing algorithms on vastly distributed edge devices

NewControl

Virtual platforms for perception and control in highly automated vehicles, based on safety by design

Computational neuroscience and bio-inspired circuits and algorithms

Low-power neuro-inspired or neuromorphic circuits and algorithms; low-power circuits and systems for neural interfacing.

History

SunRISE

Implement machine learning on encrypted data to improve security aspects of smart IoT devices

Programmable Systems for Intelligence in Automobiles

(a) fail-operational sensor-fusion framework, (b) dependable embedded E/E architectures, (c) safety compliant integration of AI approaches for object recognition, scene understanding, and decision making

RESIST

Design approach for resilient integrated electronic systems in automotive and avionics applications

Heterogeneous Inception

Developing and deploying a unified design methodology and tools for system-level design and verification of heterogeneous systems

Computing Fabric for high performance Applications

Develop an open, flexible and high performance platform by substituting heterogeneous mixed HW/SW specialized sub-systems by application specific processor arrays.