Towards a guideline for refactoring of embedded systems

The electronics in automotive systems give great possibilities. It has contributed to environmental improvements through reduced emissions and reduced fuel consumption, safety, driver assistance, and quality through better diagnostic capabilities. Automotive systems are today distributed embedded systems that consist of several nodes that communicate with each other. The increasing possibilities have led to a situation where functions that used to be stand-alone, are today dependent on several inter-connected systems which all contribute to the desired functionality. This has increased the costs and the complexity to deal with the systems. The automotive industry is adopting a new open software architecture, called AUTOSAR, that is intended to reduce the complexity. AUTOSAR also gives possibilities for coping with large product ranges and for component sharing. The introduction of AUTOSAR is an example of an architecture change without modifying the external functionality. We have chosen to call such changes system refactoring. However, if the introduction of AUTOSAR is not successfully performed, there are risks for delayed development projects, which are costly for the automotive companies. Unfortunately, existing engineering standards and literature focus mostly on new product development and less on system refactoring, and this gap needs to be filled. The goal of this research is to provide guidelines for refactoring, which provides support throughout the complete process of system architects in efforts to refactor the system. This thesis identifies the characteristics of refactoring processes. This is done by empirical studies of the drivers behind refactoring, the effects we can expect from refactoring, and the process activities and characteristics. The result can be used to create guidelines for improving the work of refactoring.