Device-Free Localization Based on CSI Fingerprints and Deep Neural Networks

Abstract

Localization is of key importance to a variety of applications. Most previous approaches require the objects to carry electronic devices, while on many occasions device-free localization are in need. This paper proposes a device-free localization method based on WiFi Channel State Information (CSI) and Deep Neural Networks (DNN). In the area covered with WiFi, human movements may cause observable variations of WiFi signals. By analyzing the CSI fingerprint patterns and modelling the dependency between CSI fingerprints and locations through deep neural networks, the proposed method is able to estimate the objects' locations according to the measured CSI fingerprints through DNN regression. To cope with the noisy WiFi channels and remove the non-contributing information, the proposed method applies Density-Based Spatial Clustering of Applications with Noise (DBSCAN) to reduce the noise in the raw CSI data, and applies Principal Component Analysis (PCA) to extract the most contributing information in the CSI data. Evaluations in two representative scenarios achieved the mean distance error of 1.08 m and 1.50 m, respectively.