Energy-Efficient Human Activity Detection in Smart Spaces

Abstract

Wireless sensor networks play an essential role in today's Internet of Things (IoT) systems. One of the most common applications is smart indoor spaces and detecting human activities in such areas. It is crucial for these systems, to collect data, analyze it, and make decisions based on the analysis. Although this is a quite well-defined pipeline of processing, the overall device energy consumption can be significant for wireless sensor systems to ensure the fidelity of data and longevity of the system. In this paper, we first discuss the energy requirements of common IoT applications using sensor networks. We focus on IoT systems for human activity detection in indoor spaces. Then, we propose a method to maximize energy efficiency for these smart spaces. Lastly, we experimentally demonstrate the effectiveness of our proposed method, both with simulation and with our real smart space deployment. Our smart environment deployment consists of a variety of sensors including ultrasonic, microwave and vibration sensors. We demonstrate that our energy efficiency method does not affect data quality, thereby maintaining the accuracy of human activity detection. Our method shows up to 30% energy efficiency improvement.