Missing Data Recovery Using Tensor Completion-Based Models for IoT-Based Air Quality Monitoring System

In IoT-based air quality monitoring system, a set of IoT devices are deployed for sensing of the air quality data at different junction of a smart city. These deployed IoT devices periodically forward the sensed data to the base station for further processing and analytics. Missing of Air Quality Index (AQI) data is very challenging issues in real-time monitoring of AQI in a smart city due to failure of IoT devices, data corruption in the wireless transmission, malfunction of sensors etc. Missing data recovery is a very fundamental issue with real-time IoT-based AQI monitoring system. To solve the missing data recovery problem, this paper has used tensor complete based data recovery models such as Bayesian Gaussian Canonical Polyadic (BGCP) decomposition, Bayesian Augmented Tensor Factorization (BAIT) and High accuracy Low Rank Tensor Completion (HaLRTC) to recovery the AQI missing data. Performance analysis of the tensor complete based data recovery models is evaluated using real time AQI dataset in terms of Root Mean Square Error and Mean Absolute Percentage Error.