Information fusion for context awareness in intelligent environments

Abstract

The development of intelligent environments requires handling of data perceived from users, received from environments and gathered from objects. Such data is often used to implement machine learning tasks in order to predict actions or to anticipate needs and wills, as well as to provide additional context in applications. Thus, it is often needed to perform operations upon collected data, such as pre-processing, information fusion of sensor data, and manage models from machine learning. These machine learning models may have impact on the performance of platforms and systems used to obtain intelligent environments. In this paper, it is addressed the issue of the development of middleware for intelligent systems, using techniques from information fusion and machine learning that provide context awareness and reduce the impact of information acquisition on both storage and energy efficiency. This discussion is presented in the context of PHESS, a project to ensure energetic sustainability, based on intelligent agents and multi-agent systems, where these techniques are applied.