A memory efficient algorithm for network reliability

We combine the augmented ordered binary decision diagram (OBDD-A) with the use of boundary sets to create a method for computing the exact K-terminal or all-terminal reliability of an undirected network with failed edges and perfect vertices. We present the results of implementing this algorithm and show that the execution time is comparable with the state of the art and the space requirement is greatly reduced. Indeed the space remains constant when networks increase in size but maintain their structure and maximum boundary set size; with the same amount of memory used for computing a 3 × 12 and a 3 × 1000 grid network.