Master's thesis

Title: Decision Diagrams for the Efficient Representation of Automotive Configurations available: here

This thesis investigates how the extremely large space of automotive product configurations can be represented and validated efficiently. It extends the decision-diagram framework JINC to support Multi-valued Decision Diagrams (MDDs) with a variable number of branches.

Several MDD variants are implemented and compared, including fully reduced, quasi-reduced, and sparse representations. In addition, a new branch compression technique based on run-length encoding is introduced to significantly reduce memory usage.
The evaluation shows that variable ordering and constraint ordering are the key factors affecting performance and memory consumption.
Overall, the results demonstrate that MDDs are an effective approach for representing and validating complex automotive configuration spaces.