Node failure search via the multiscale averaging method

Abstract

In this paper, we propose a search algorithm called the Multiscale Averaging Method (MAM) to locate the level of an error node in a branching network. Complications due to the additional connections (AC) on the network cause the detection probability (P) to drop with increasing AC until 0.25 of the total possible AC. Interestingly, however, beyond 0.25, P increases with AC. Hence, AC complications which can be interpreted as noise can improve the probability of failure detection. We suggest that this behavior is reminiscent of a type of noise-aided effect called dithering or the stochastic resonance.