Fault tolerance in fog-based Social Internet of Things

Abstract

The Social Internet of Things (SIoT) is a glimpse into the future for self-establishment of inter-things social relationships that guarantee object-to-object transactions. This scenario adopted an intelligent strategy by the trust characteristics and typology of leveraging objects to eradicate the interference of defective nodes. In addition, SIoT networks are error-prone due to heterogeneity and potential resource constraints on the devices, which necessitates the inclusion of a mechanism to enhance the lifetime and dependability of devices. Therefore, fault tolerance mechanisms are indispensable to ensure optimal performance through fault detection and recovery. Although fog architecture and SIoT are two stand-alone paradigms, we extend fog competence into SIoT to empower resource-poor objects for handling intensive tasks. Hence a fault-tolerant model based on a clustering algorithm for fault detection and retrieval is presented. In the proposed scheme, the Markov chains model all faults diagnosed by the fog node. In the recovery stage, the status of the nodes is determined based on the type of faults and the assigned threshold to make a substitution by hot standby nodes. The four performance criteria, namely the reliability, availability, probability density function, the average time to failure, and their mathematical equations were developed on the Markov model. Furthermore, the achievements were demonstrated through either theoretical analysis or simulation results. Then, the proposed scheme increases the number of live nodes, improves the detection accuracy in forwarded requests, lowers the number of invalid requests, and enhances availability and reliability compared to networks with analogous properties.