An Energy-efficient and Lightweight Indoor Localization System for Internet-of-Things (IoT) Environments

Abstract

Each and every spatial point in an indoor space has its own distinct and stable fingerprint, which arises owing to the distortion of the magnetic field induced by the surrounding steel and iron structures. This phenomenon makes many indoor positioning techniques rely on the magnetic field as an important source of localization. Most of the existing studies, however, have leveraged smartphones that have a relatively high computational power and many sensors. Thus, their algorithmic complexity is usually high, especially for commercial location-based services. In this paper, we present an energy-efficient and lightweight system that utilizes the magnetic field for indoor positioning in Internet of Things (IoT) environments. We propose a new hardware design of an IoT device that has a BLE interface and two sensors (magnetometer and accelerometer), with the lifetime of one year when using a coin-size battery. We further propose an augmented particle filter framework that features a robust motion model and a localization heuristic with small sensory data. The prototype-based evaluation shows that the proposed system achieves a median accuracy of 1.62 m for an office building, while exhibiting low computational complexity and high energy efficiency.