Calibration-free wireless indoor localization (CAFLOC)

Abstract

Indoor localization is a very promising field in the era of Internet of things (IoT) and has a large number of potential applications. Due to the popularity of mobile devices in recent years, using the Received Signal Strength (RSS) of Wi-Fi signals and a fingerprint database (a.k.a radio map) for indoor localization becomes quite attractive. One obstacle is that the RSS values on the reference device (the one used to build the radio map) and the target device (the one whose location needs to be determined) are not identical, resulting in localization errors. To reduce the errors, a costly and time consuming calibration process is often used. In this paper, we propose a novel CAlibration Free LOCalization (CAFLOC) approach that utilizes relative RSS information to locate target devices and does not require any calibration. Our method relies on the linear relationship between the RSS values on the reference and target devices, which has been reported by many papers in the literature and also verified in our LAB. We first show mathematically why such a linear relationship exists. We then present CAFLOC and prove that in ideal situations, it is able to precisely identify locations without errors. To verify its performance in real-world scenarios, we run extensive localization tests on CAFLOC and the Nearest Neighbor (NN) approach. Our results consistently show that CAFLOC is much more accurate than NN.