Evaluation and Comparison of Performance Analysis Methods for Distributed Embedded Systems
In this thesis we evaluate and compare a number of system level performance analysis methods for distributed embedded systems. We discuss different criteria for their classification and comparison and evaluate some concrete performance analysis techniques with respect to these criteria. In particular, we examine the modeling power and usability of various approaches, apply the performance analysis methods to a set of benchmark systems and compare the obtained results. We show that there are important differences between methods in terms of modeling effort and accuracy by highlighting some modeling difficulties and analyzing pitfalls of the different formal approaches. Further we present an extendible open-source library for performance simulation of distributed embedded systems based on SystemC and compare the hard performance bounds provided by formal analysis methods with the performance estimations obtained by simulation.