Compressing Configuration Data for Memory Limited Devices

The paper introduces a new type of compression for decision diagram data structures, such as BDDs, MDDs and AOMDDs. The compression takes advantage of repeated substructures within the decision diagram in order to lessen redundancy beyond what is possible using simple subfunction sharing. The resulting compressed data structure allows traversal of the original decision diagram with no significant overhead. Specifically it allows the efficient computation of valid domains, that is, the assignments for each encoded variable that can participate in a solution, which is critical when the decision diagram is used to support an interactive configurator. We relate these results to applications for interactively configurable memory limited devices and give empirical results on the amount of saved space for a wide variety of instances.