The Information Propagation Method for Efficient Network Reliability Analysis

In this paper, we present an ordered message-passing algorithm for propagating information related to system component conditions and dependency between nodes in the target graph network. This tool, called Information Propagation Method, exploits the structure of Directed Acyclic Graphs to assess the reliability of a network. The key advantage of this analytical approach is the computation efficiency obtained by maintaining a priority set of nodes to be updated in each step, every node in the graph is processed as infrequently as possible. Whilst the algorithm converges to an exact solution in simpler graph structures, the solution in more complex graphs is approximated due to the presence of the so-called diamond subgraphs. Diamond subgraphs are identifying solved and recursively decomposed by the algorithm allowing the exploration of complex component dynamics as well as system redundancy strategies. The algorithm is verified against the solution obtained via Monte Carlo simulation. Finally, we discuss a case study example showing the applicability and advantages of the proposed algorithm.