Activity Prediction in a Smart Environment Using Bayesian Network and Multi-Agent System

Technology in devices such as SmartPhones is transforming people's lives. It's also being incorporated into homes in various ways, and hence Smart Environments are becoming more accessible and real. This work focuses on studying and testing a Bayesian Network for anticipating inhabitant interactions with a Smart Environment. It also uses Multi-Agent Systems to model the environment's functionalities allowing for expandability. As a case study, Multi-Agent System and Bayesian Network run in a Raspberry Pi embedded platform. In addition, ESP8266 embedded platform is used to develop automated devices connected to the local network, enabling the achievement of a Smart Environment. The case study shows that user's interactions with the environment is reduced by more than 50% actuating over the Wi-Fi with a standard delay around one second.