Performance Analysis of Distributed Real - Time Embedded Systems Master thesis

The design of a distributed real-time embedded system is a difficult job. The hardware and software parts are often designed sequentially, leading to over-conservative and expensive systems. A more reliable and optimal system is obtained by introducing performance analysis in the early design phases. This analysis is performed with models designed in an ad-hoc way. We propose a method which uses models to analyse distributed real-time embedded systems that capture both functional and timing properties, in an early design phase. The models are based on SHE (Software/Hardware Engineering), a system-level design methodology based on the formal modelling language POOSL (Parallel Object-Oriented Specification Language), and on the fast execution engine Rotalumis. The modelling method is based on the Y-chart scheme concepts and involves specification of the environment, the application, the architecture and the mapping between them. This thesis presents modelling patterns for common input/output devices, real-time tasks and platform resources. With these patterns a model of a distributed real-time embedded system can be build conveniently. The patterns are validated by means of a realistic case study which is an in-car navigation system. The outcome of the performance analysis is compared with the outcome of Modular Performance Analysis (MPA), a method based on worst-case execution time analysis. The comparison shows that the proposed method produces performance numbers that approximate the worst case execution times as opposed to MPA which calculates the WCET and is sometimes over-conservative. The proposed method effectively captures the behaviour of both soft and firm real-time embedded systems. ii Acknowledgements I hereby thank the people from the TU/e ICS/ES department for giving me the opportunity of making this thesis. I would like to thank Jeroen Voeten, my supervisor, for giving me the opportunity to develop my own ideas. My special thanks go to Oana Florescu for coaching me during my research work. From the early stages until the final version, she was always able to help me in structuring my thesis and gave me helpful feedback to improve my work. I also want to thank Marcel Verhoef, a member of the Boderc project carried out at Embedded System Institute for providing me with the In-Car navigation system case study. Finally, I wish to express my thanks to my family and friends, who have supported me during my time at TU/e and my graduation. Special thanks go to my parents for giving me the chance to continue my studies at …