Optimal Product Line Architectures for the Automotive Industry

Abstract

The creation of product line architectures is a difficult and complex task. The resulting architectures must support the required system variabilities as well as further quality attributes. In the automotive domain, product lines of software-intensive system models have a great diversity of products, which leads to vast design spaces. Finding optimal product line architectures as part of the system design process requires the consideration of a variety of trade-offs. In practice, this challenge cannot be solved manually for all but the smallest problems, therefore an automated solution is required. Our contribution is the generation of a sound mathematical formalization of the problem. This formalization makes the product line optimization problem accessible to various established multi-objective optimization techniques. The applicability of the chosen approach is shown by means of applying a commercial tool for multi-criteria decision making.