Ranging and fingerprinting‐based indoor Wi‐Fi localisation using channel state information

Abstract

In this study, a hybrid indoor localisation scheme based on ranging and fingerprinting using the calibrated channel state information is presented. In ranging, regarding the serious indoor multipath effect, the authors propose a ranging scheme based on channel impulse response to improve the ranging accuracy and stability of the existing ranging scheme using received signal strength. In fingerprint extraction, they utilise phase difference between antennae to generate fingerprint. The phase difference is more stable and better represents a specific location. Different from the traditional fingerprint-based scheme including establishing fingerprint database and matching all the fingerprints, they use multi-layer perceptron to learn a classification model by training on a fingerprint dataset and the new fingerprints are feed to the model to predict the corresponding locations. Their scheme can significantly reduce the computational overhead by the introduction of the distance between transceivers. Finally, they conduct and validate the proposed scheme in two typical indoor environments by extensive experiments. The result shows that the localisation accuracy can be significantly enhanced using their scheme.