Building Data-Aware and Energy-Efficient Smart Spaces

Abstract

With the increase in the dependency of our life on technology and data, smart spaces have become integral in providing an environment for data collection, analysis, and machine responses. This paper discusses the current research in this field and the challenges that arise in the execution of these smart spaces. We address the major challenges of hardware design, data analysis, and energy efficiency in a new data aware smart environment that collects time-stamped data for position, movement, temperature, and vibration sensors. Data collected from these sensors is used to achieve energy efficiency, for real time localization in conjunction with machine learning mechanisms to analyze human activities. We evaluate six different machine learning algorithms for human activity detection task, on a data set collected in our laboratory. Results show high classification performance for all methods giving up to 99.95% classification accuracy. We also implemented energy-efficiency measures, leading to up to 30% energy efficiency improvement on top of our initial design. This ambient environment, along with data analytics and improved energy efficiency, provides information regarding the occupancy and behavior of people within its range. Spaces such as conference rooms, common areas such as libraries, classrooms, and even public spaces such as public transport can benefit from our design. Our system avoids privacy issues by using no audio/visual devices. This system thus provides an insight into smart spaces, their current trends, and what future direction research such as ours would lead them to.