On the continuous inter-group interaction speed improvement in large-scale agile software development for embedded software

Abstract: An open issue in industry is how to scale-up Agile Software Development in large companies developing embedded systems. Intra-team interaction is well supported by established Agile practices, while how to handle the feedback for the Agile team from other teams and other parts of the organization is still unexplored. The adaptation of Agile Software Development (ASD) to em- bedded software, driven by an overall physical product development process, brings challenges in terms of dependencies on parallel engineering and manage- ment activities. Such interactions slow down the development, hindering the achievement of business goals based on speed, such as a short time to market, the quick replication of products of a product-line, or the reaction time for a change request. This thesis is based on a 2-year-long action research strategy, collecting data through interviews, a survey (questionnaire) and focus groups in 3 large companies developing embedded software. Through a combination of Grounded Theory (qualitative) and statistical (quantitative) analysis, to- gether with continuous empirical validation of results, we have developed a framework for monitoring and improving interaction speed. The purpose of such a framework is to identify and eliminate speed waste caused by interac- tion issues. The final goal is to improve the feedback between the development team and specifically selected parts of the organization that, we show, are crit- ical. The framework involves the recognition of visible symptoms leading to the underlying causes (factors) and the application of a set of practices for the mitigation of such factors. Such practices represent steps and activities to be implemented for software process improvement (SPI).