Model-based product line engineering for embedded real-time systems

Product Line Engineering (PLE) has attracted attention in the last two decades due to its promising capabilities to reduce system life-cycle costs and time to market through the reuse of requirements and components. Current PLE methods, however, mainly focus on the software aspects and are called Software Product Line Engineering (SPLE). While SPLE is important in the systems domain, Systems Engineering is a much broader discipline; it includes the consideration of hardware aspects (mechanical and electrical engineering) as well as the overall integration of hardware and software components.This thesis proposes a novel and practical method to support a development of Model-based Product Line Engineering (MBPLE) for Embedded Real-Time Systems (RTS) for both Domain Engineering and Application Engineering processes. It consists of four main contributions: (i) a variability modelling framework that provides capability to capture variabilities with Systems Engineering disciplines. (ii) a Domain Engineering framework to combine proposed variability model (introduced in (i)) with system models to build system reusable artefacts. (iii) a Product Line Systems Engineering Process that consumes legacy system models to semi-automate the generation of variability model and variability dependency (achieved through formalising dependencies existed in the variability model, the functional artefacts and between the two); and reduces configuration efforts of variability model by a proposed algorithm (grounded in formal notation and mathematical framework, which promises trustworthiness and verifiability of the reduced variability model). (iv) an approach that extends the UML profile for ROSETTA (UPR) for proposed variability framework to facilitate trade-off analysis during Application Engineering process for target system generation.To verify the feasibility and evaluate the effectiveness of the proposed MBPLE for Embedded RTS, it is further applied to case studies in different engineering domains including Engine Control Systems, Aircraft Systems, Logistics Systems, Automotive Systems and Remote Health Monitoring Systems at different levels of complexity. Subjected to system model availability, a reduction of 14%–50% in the number of variation points is demonstrated in the case studies.