A Monitoring System and Faults Prediction for Internet of Things System

With the emerging Internet of Things (IoT), the real-time applications are increasingly deployed on IoT systems using Multi Processing System On Chip (MPSoC). Real-time applications have strict temporal constraints, and are not fault-tolerant. Any defective IoT device can lead to serious faults such as data loss or even application failure. It is possible to predict the defective IoT device using a predictive model to prevent faults. Data on CPU load, consumption and thermal state of devices are correlated with the state of the device. In this paper, we show how to classify the state of MPSoC IoT Devices using a decision tree with ID3 algorithm as a split heuristic based on hardware measurement features.