Introduction to Model-based Reliability Evaluation of Wireless Sensor Networks

A high level of reliability is a significant requirement for using wireless sensor networks in industrial environments. Model-based evaluation is usually applied in conventional systems to estimate the reliability. In contrast, for analyzing sensor networks, these methods are hardly tested and proven due to the unique properties of that kind of network. This paper presents a first model-based approach to quantitatively assess the reliability of sensor networks. Based on an abstract model of a sensor network, the degree of detail with respect to the characteristic aspects of sensor networks is increased in a stepwise manner. If network topologies are taken into account, analytical methods fail and the assessment of reliability measures has to be done numerically. Finally, a Monte Carlo Simulation is conducted to also cover dynamics in the topology. In conclusion, this paper shows that, by knowledge of link probabilities and lifetime distributions of single sensor nodes, model-based analysis may be used to estimate reliability measures of sensor networks.