An Improved Constraint Ordering Heuristics for Compiling Configuration Problems

This paper is a case study on generating BDDs (binary decision diagrams) for propositional encodings of industrial configuration problems. As a testbed we use product configuration formulas arising in the automotive industry. Our main contribution is the introduction of a new improved constraint ordering heuristics incorporating structure-specific knowledge of the problem at hand. With the help of this constraint ordering, we were able to compile all formulas of our testbed to BDDs which was not possible with an arbitrary constraint order.