Improved Energy Efficiency of Indoor Positioning Systems by Adaptive Sampling and Smart Post-processing of Sensor Data

Abstract

In this paper, a method for minimizing power consumption of indoor positioning and tracking systems is proposed. Most Dead-Reckoning (DR) systems nowadays available are based on an Inertial Measurement Unit (IMU) operating at a relatively high sampling frequency of more than 100 Hz in order to track the typical human motion with a sufficient resolution. In this work, a method is proposed which allows to reduce the sampling frequency for sensor data by a factor of more than 10. The inaccuracies and fluctuations associated with the low sampling rate are compensated by applying a correction algorithm that takes the geometry of the underlying environment as well as the previous motion direction into account. A similar concept for sampling rate reduction and post-processing will also be applied to the data of an air pressure sensor. This sensor type can be used for movement detection in vertical direction. The adjustment of the sampling frequency is done adaptively based on the movement of the person subject to track. It will be shown that by using the proposed methods the accuracy of position tracking is comparable to that of an Indoor Positioning System (IPS) operating at 200 Hz whereas the data rate and therewith the power consumption is reduced by a factor of more than 10. This is verified by measurements using an experimental setup based on low-cost Bluetooth Low Energy (BLE) modules.