Vulnerability of Average Consensus Algorithm: Impact of Stochastic Communication Breakdowns on Generalized Metropolis-Hastings Weight Design

The fault tolerance is an important design issue of many real-life applications based on the wireless sensor networks. However, the implementation of protective mechanisms results in increased overall energy consumption, making its usage not optimal. This paper focuses on the fault tolerance of the average consensus algorithm with the generalized Metropolis-Hastings weight design with a modified mixing parameter and without any protection. The robustness of the algorithm is examined using an MSE-metric for four various initial configurations and the obtained results are compared with concurrent weight designs for various failure probabilities.