Diagnosing tree-structured systems Original Research Article

This paper introduces the TREE/TREE∗ algorithm for computing minimal diagnoses for tree-structured systems. Diagnoses are computed by descending into the tree, enumerating the input combinations that might be responsible for a given incorrect observation, and combining the diagnoses for the subtrees generating these inputs into diagnoses for the whole system. Algorithm TREE diagnoses systems containing functional components and algorithm TREE∗ diagnoses more general constraint-based components. We prove soundness and correctness of the algorithms and show experimental results that indicate that they compare favorably to Reiterʹs hitting-set-based algorithm and El Fattah and Dechterʹs SAB. Extensions of the algorithms such as use of fault modes are discussed.