Detecting Change in Activity or Identifying Failure sensors in IOT Using Frequent Count Activity Matrix (FCAM) Algorithm
The Internet of Things (IoT) and Machine Learning are the leading cutting-edge technology with Home Automation or Smart Homes as one of the application and their combination can do wonders with sensors. However, the sensors associated with interconnected devices in Smart Homes often imparts erroneous results due to deviations caused by external factors that in turn affects the performance of these devices. Thus, it becomes necessary to detect deviations in the performance of devices and identify the faulty sensors. The proposed work contributes to two stages, first stage being used to detect the change of activities in Home Automation using the design of Frequent Count Activity Matrix (FCAM) and in second stage, it confirms the missing sensor epochs is due to change in activity or sensor failure using sensor frequent count array. The FCAM matrix has been constructed using real world data set known as a learning matrix that intends to compute minimum and maximum threshold values for each activity from the learning matrix. The training activity data set uses the learning matrix and sensor frequent count array to detect the changes in activity. The execution time has been observed in centi-milliseconds that improves the performance of the algorithm using Python language and Data Visualization Libraries such as Pandas and Matplotlib.