Section: New Results

Traceability of Concerns in Large Software Systems

In 2015, we obtained new results in the domain of the analysis of large software systems. The purpose is to be able to deal the complexity of such systems by slicing them depending on different concerns. The slicing enables to gain a view and a better understanding on how the concern evolves over time and through the different refinement layers of the software system. For that, we present a systematic approach based on model driven engineering and basic models of software components, in order to better manage software complexity and traceability of functional and non-functional requirements. We provide in particular three major contributions. First, we provide an integrated set of meta-models for describing the concerns of software requirements, software components, and traceability between the concerns and software components. By providing an abstract model, we are independent of any implementation and thus allow existing approaches relying on that model to expand their support. With the second contribution, we propose a formal support of our model to allow formal verification. We focus on temporal property verification. For this, our design model is translated into timed automata for which we can apply a timed model checker. Instead of using temporal logic, which is difficult to handle by non-experts, we use patterns of temporal properties. For each pattern, we propose timed automata that can be applied directly into a timed model checking tool. These timed automata are seen as observers or watch dogs that check the system under observation. Finally, with the last contribution, we propose a software component-based development and verification approach, called SARA, and included in V-lifecycle widely used in the railway domain. These contributions have been validated with case studies from the domain of railway control systems especially for the new European train control system ERTMS/ETCS. These results contribute to our objective on self-optimizing software systems (see Section  3.3 ) and are part of the PhD thesis by Marc Sango [13] .