Investigation of the structure of a networked system

Practical networked systems are composed by components, which can be modelled as connections between unfaultable nodes, subjected to failure and repair events. Generally, the network is characterized by deterministic and/or stochastic dependences among the states of its components [1]. Different algorithms have been proposed for the connectivity analysis of a network; an algorithm based on Cellular Automata, which does not require the evaluation of the system structure function(s), is applied in order to evaluate the state of all “end-user” nodes (connected/not connected to the “source” node(s)) for each system configuration [2]. An “indirect” approach, which does not suffer any limitation from the introduction of dependences among events, is applied in order to sample by Montecarlo techniques the random walk of the system among its configurations [3], [4]. In this paper, importance sampling techniques are applied in order to manage rare events (generally due to the different orders of magnitude of the failure and repair frequencies and to the high number of redundant paths). The failure and repair of all components are forced to occur with the same probabilities; a system transition due to the failure of a component is forced to occur certainly within its mission time [5]. The proposed approach is applied for the analysis of the availability performances of a networked system fully characterized, in order to verify its effectiveness.