The need for increased computing capability and more diverse hardware with its evermore complex topologies continues to grow. The use of multicore processors, which is quite established in the desktop computers, is gaining importance in the embedded systems for industrial applications. Because of the complexity of the environment and the special requirements for these applications, the embedded multicore processors require optimised software architectures and the corresponding design methods. Performance optimisation is often based on the detailed knowledge of program behaviour. One such option to obtain information of program behaviour is software tracing, which forms the core of this thesis work. Software tracing is being used for a while now to identify faults, anomalies and bottlenecks in sequential as well as parallel applications. Tracing is not only limited to collecting and recording the event data from applications, but also to utilise the trace data in a meaningful way for further analysis and visualization. It is also not trivial to classify or analyse the trace data, especially when the amount of trace data exceed a million of events per second. For this reason, there is a need to create a high level model of trace data. In this thesis, we look at the theoretical model which describes the classification of trace events and also the relation between the events in a trace. The context for performing analysis on trace events is also discussed in detail. Based on the model, we have also created a prototype tool using Eclipse Modelling Framework (EMF) which performs the analysis on traces from real-time applications.