Abstract

Technology at home is on the rise and the place for smart home solutions is growing.

Smart homes are built by adding intelligent and adaptive behavior to home automation systems. Smart homes are envisioned to improve comfort, control, safety and enhance the overall experience. One way to achieve this goal is detecting anomalous events or actions. This means searching for rare events to occur that could end up in a situation of concern or interest. In this work we combine the Internet of Things (IoT) technology with Artificial Intelligence (AI) to deliver a solution which ingests data from sensors, analyzes it, responds to any changes and it would eventually predict the possible flaws. We employed IoT sensors that are able to monitor temperature, humidity, CO2, light levels and energy consumptions. Preprocessing and visualization of data, AI model development, and model assessment is done in Python by using proper AI libraries and frameworks. Both Machine Learning (ML) and Deep Learning (DL) models have been developed and trained.

We evaluated the metric and the performances of the two models and carried out a predictive analysis on the best one. Accurate predictions are able to address the issue of detecting anomalous events or actions.