A Method of Combining HSSE-tree and Binary Label to Compute All Minimal Hitting Sets

Abstract

Computing all minimal hitting sets (MHSs) is a key step of model-based diagnosis. A novel method to compute all MHSs called Binary-label HSSE is put forward, which combines HSSE-tree and binary label. In the method, binary digits is used to mark the real elements of the nodes, and effective pruning and expanding strategies are used to avoid the main problem of HSSE-tree, the explosive growth of the expanded nodes and supersets of MHSs along with the dimension of the problems. Additionally, computing between binary digits can avoid the traverse of every element in a node when judging whether the node is a MHS, which also contributes to the significant decrease of the run time. At last, the data structure of Binary-label HSSE is changed to dynamic array from dynamic linked list, which further decreases the run time of the method. Simulation results show that Binary-label HSSE method costs much less space and time than HSSE-tree method.