N-Version Programming for Enhancing Fault Tolerance in Fog-based IoT Systems

With the increase in the abundance and prominence of fog-based systems comes the increase in demand for smarter devices. This can be quite challenging since fog-based IoT systems need to adapt in the event of a sudden deterioration in the level of service they offer due to hardware or software fluctuations. Fog-based IoT systems need to become fault-tolerant in order to ensure the delivery of secure, reliable, robust, and dynamic services while addressing unexpected changes that may occur in terms of both hardware and software. To achieve such fault-tolerance, however, it is imperative to accurately define and identify the differences between errors, faults, and failures that may exist within fog-based environments. In this paper, we propose a solution to this problem and introduce an N-version anomaly-based Fault Detection (NvABFD) technique used for enhancing the fault tolerance of fog-based systems. Using NvABFD, it is possible to identify data anomalies, errors, faults, and failures that may occur in fog-based environments in near real time. We tested the NvABFD technique in a simulated patient monitoring system using the MobiAct dataset. Our results show an accuracy of similar to 99.9% in anomaly, error, and fault detection indicating that this technique may enhance fault-tolerance in a fog-based system by accurately identifying anomalies, errors, and faults as they occur.