Reliability Modeling for Industrial Edge Computing Systems

Monitoring of extended industrial facilities in hard-to-reach places is a great challenge for which solutions based on wireless sensor networks are proposed. Edge computing paradigm allows to improve the effectiveness of monitoring systems. However, in order to fully utilize this potential, it is necessary to develop tools for analyzing the reliability of these systems. The paper proposes a method for assessing the reliability of specialized Edge computing systems. We consider unreliable sensors and assume that a sensor is functioning with a given probability. Previously, sensors have been organized into clusters according to application needs. Cluster members communicate with the outside world through a specialized node called a gateway. Therefore, the cluster topology is a star with a gateway in the center, and hence the system topology is star interconnection over a highly reliable common bus. The gateway performs a number of functions, such as processing the monitoring data in some way before further transmission. To describe the gateway behavior, we use a continuous time Markov chain. Thus, we have developed a model for the reliability of the system, taking into account the topology and dynamics of the network.