Abstract
•Reliability of Internet of Things (IoT) systems considering cross-layer dependence•A dynamic Bayesian network (DBN)-based model is suggested•Reliability updates using the Dempster-Shafer's theory•Case studies on smart grid and multi-state smart fire protection systems
The Internet of Things (IoT) technology is widely applied across various critical fields due to its intelligent and convenient features. Consequently, reliability assessment of these IoT-based systems has gained increasing attention. However, the existing reliability models often fail to fully address the collaborative functionality of different IoT layers. To fill this research gap, we propose a dynamic Bayesian network (DBN)-based model for reliability evaluation and updating in IoT systems, explicitly capturing interdependencies across the perception, communication, support and application layers. Particularly, reliability analysis of the IoT system is conducted using DBN inference. Observation data during the service are merged using Dempster-Shafer theory and an improved evidence theory to facilitate reliability updates. The effectiveness and broad applicability of the proposed approach are validated through case studies on a smart grid system and a multi-state smart fire protection system. Through addressing cross-layer dependencies, the reliability results assessed using the proposed method can enhance decision making in maintenance planning activities and reduce operational risk in various IoT systems.