Performance Analysis and Component-Based Design of Real-Time Embedded Systems

Embedded systems are getting increasingly parallel, distributed and networked, i.e. they are comprised of many cooperating individual components. Examples of these systems are networks of sensors and actuators, MPSoC (multiprocessors on a chip), multicore architectures, embedded systems in automotive and avionics. It appears that there are several fundamental problems that make the design of distributed embedded systems difficult:

  • handling non-functional and resource constraints,
  • design under multiple conflicting criteria,
  • trade-off between average performance and predictability,
  • component-based design strategies.

It appears that conventional computer science and engineering methods are at their limits. The scaling up in distributivity and size asks for a paradigm shift in models and methods.

Modularity is a key requirement for any performance analysis and design method as distributed systems are heterogeneous in terms of (a) the underlying execution platform, (b) the diverse applications running concurrently and (c) the different scheduling and arbitration policies used. In this context, we are investigating analytic methods based on network calculus and real-time calculus that allow worst case and best case analysis results of distributed embedded systems.

More information can be found on the home pages of the specific projects we are and have been involved in:

Related mathematical tools and high-level compilation frameworks:

Some relevant publications on our work since 2000: Publications on Modular Performance Analysis, Publications on Low Power Design.