Scalable Product-Based Consistency Analysis of Cyber-Physical Product Lines

Configurable cyber-physical systems can be managed as cyber-physical product lines to deal with the growing complexity and variability of modern industrial systems. Product lines are usually divided into a problem and a solution space. The problem space comprises the theoretical variability of the product line in the form of configurations while the solution space comprises hardware and software artifacts that realize these configurations into products. Ensuring the consistency between the problem and solution space of a product lines is essential, as certain valid configurations may lead to non-realizable products, because of a mismatch between the modeled variability in the problem space and the realizable variability in the solution space. The state-of-the-art product-based approach for analyzing the consistency of cyber-physical product lines struggles to scale with increasing system size as it analyzes each configuration naively. This thesis aims to improve the scalability of the
product-based approach through three enhanced analysis strategies: A repository-based analysis strategy that reuses results from previously analyzed configurations,
... mehr
a parallel analysis strategy that distributes configuration analysis across computation units, and a parallel repository-based analysis strategy that combines reuse of
previous analysis results and parallelism. We evaluate the three analysis strategies using the evolution history of a well-known case study from the automotive industry using the product-based approach as a baseline for comparison. The results show that, in terms of run-time, each analysis strategy is at least 4.74 times faster
than the product-based approach in each version of the case study. However, the results also show that none of the strategies scales linearly with increasing numbers
of configurations in a product line.