Multi-layer Fault Localization Using Probabilistic Inference in Bipartite Dependency Graphs

For the purpose of fault diagnosis, communication sys- tems are frequently modeled in a layered fashion imitating the layered architecture of the modeled system. The layered model represents relationships between services, protocols, and func- tions offered between neighboring protocol layers. In a given layer, an end-to-end service between two hosts may be provided using multiple hop-to-hop services offered in this layer between two hosts on the end-to-end path. When an end-to-end service fails or experiences performance problems it is critical to effi- ciently find the responsible hop-to-hop services. Dependencies be- tween end-to-end and hop-to-hop services form a bipartite graph whose structure depends on the network topology in the corre- sponding protocol layer. To represent the uncertainty in the de- pendency graph, probabilities are assigned to its nodes and links. Finding the most probable explanation (MPE) of the observed symptoms in the probabilistic dependency graph is NP-hard. We transform the bipartite dependency graph to a belief network and investigate several algorithms for computing MPE such as bucket tree elimination and two approximations based on Pearl's itera- tive algorithms. We also introduce a novel algorithm using an it- erative hypothesis update. These algorithms are implemented in Java and their performance and accuracy are evaluated through extensive simulation study.