Evaluate Voting System Reliability Using the Monte Carlo simulation and Artificial Neural Network

Abstract

The threshold voting system (TVS) is a generalization of k-out-of-n systems. It is widely used in human organization systems, technical decision-making systems, fault-tolerant systems, mutual exclusion in distributed systems, and replicated databases. The TVS comprises of n units, each of which provides a binary decision (0 or 1), or abstains from voting. The system output is 1 if the cumulative weight of all 1-opting units is at least a pre-specified fraction tau of the cumulative weight of all non-abstaining units. Otherwise, the system output is 0. In this study, an intuitive Monte Carlo simulation (MCS) was first developed to estimate the TVS reliability value. Then a new artificial neural network (called MCS-ANN) and a response surface methodology (called MCS-RSM) with the box-Behnken design (BBD) were created to find the approximated reliability function from the reliability estimated by MCS. The effectiveness of these two approaches were also compared using a benchmark TVS.