Abstract:
Smart IoT systems develop with the combination of cloud and IoT-based solutions.
Such systems assist by automating manual systems of everyday life and providing
comfort, security, privacy, and monitoring of intelligent devices more effectively.
However, these systems should provide failure-free execution of actions and automation, which is impossible. Faults may occur due to failures that may surface
due to hardware, frmware, or software. It introduces a factor of unreliability in
smart IoT systems and increases software and hardware-level development costs.
There are various devices and they report diverse and distinguished statuses. We
propose predictive systems for faults, failures and angle tracking in which we do
experiments for the different nature of relevant data obtained. Our proposed solutions frst obtain and process data for faults prediction, failure rate prediction and
angle tracking forecast and save them into the cloud for historical analysis. We
propose techniques for obtaining and processing data for the proposed solutions.
In the experiments, we prepare models using deep learning techniques to predict
the next failures rate per attempt, next possible faults and forecast angle tracking
using Bi-directional Long Short-Term Memory (Bi-LSTM) and Gated Recurrent
Unit (GRU). Our systems help predict the reliability behavior of devices using
either its failure rate per attempts or faults prediction that would support a more
comprehensive failure-free working of these innovative IoT systems. Our predictive systems will help service providers extend better systems’ serviceability and
assist management in making better and timely decisions. Early prediction helps
consumers to get alerts and take necessary actions on them. We use mean squared
error (MSE) and root means squared error (RMSE) as a performance evaluation
measure in failure rate prediction and angle tracking forecast experiments. We
use accuracy, F1 score, recall and precision as performance evaluation metrics in
faults prediction.
