A Predictive Framework for Location-Aware Resource Management in Smart Homes

Abstract

Rapid advances in a wide range of wireless access and networking technologies, along with ubiquitous computing and communications, have set the stage for the development of smart environments (for example, smart homes or offices). Context awareness is perhaps the most salient feature in such intelligent (indoor) environments. Examples of contexts include the location and activities of the inhabitants among others. In this paper, we develop a predictive framework for location-aware resource optimization in smart homes based on the hypothesis that the mobility of an inhabitant creates an uncertainty of his location. With the help of information theory, the proposed framework is shown to minimize this uncertainty through optimal learning and prediction of the inhabitant's movement (location) profiles captured in the symbolic domain. The concept of the asymptotic equipartition property (AEP) is also used to predict the inhabitant's most likely routes (or path-segments) with a high degree of accuracy. Successful predictions help in several ways, such as automated device control and proactive reservation of resources (for example, electrical energy and other utilities or scarce wireless bandwidth for mobile multimedia applications) along the inhabitant's most probable locations and routes. The goal here is to minimize the maintenance and operations cost of the smart home, as well as to provide better comfort to the inhabitants. Experimental results from a typical smart home floor plan corroborate the prediction success and significant reduction in daily energy consumption and manual operations of devices.