The SoC-Trace project aims to make it easier to evaluate and debug embedded applications. These capabilities will become increasingly important to ensure the reliable, effective operation of embedded systems as hardware and software components become ever more complex. This scale factor stands to make existing systems rapidly obsolete. The main obstacle will be dealing with the large number of cores and parallel execution paths. One option for overcoming this hurdle is to use execution traces--but that would require having the right trace analysis methods and software available.
The project will develop new trace analysis methods using data processing techniques like data mining, visualization, data aggregation, and probabilistic analysis. The methods will automatically spot anomalies and analyze the complicated correlations and dependencies among the many hardware and software components of embedded applications. The project team will use the following methods to build on the current state-of-the-art: * Filtering, structuring, and abstraction to reduce the number of traces to analyze; * Data aggregation; * More effective visualization.