Multipath-Based CSI Fingerprinting Localization With A Machine Learning Approach

In this paper, we propose a multipath-based channel state information (CSI) fingerprinting positioning scheme in outdoor environment with a machine learning approach. In this scheme, multiple distinguishable paths are exploited to extract discriminative features of one location automatically. Meanwhile, we design a hierarchical architecture to train and position in two stages which improves the accuracy of localization effectively. Experimental results show that the location can be estimated with an accuracy about one meter.