Channel State Reconstruction Using Multilevel Discrete Wavelet Transform for Improved Fingerprinting-Based Indoor Localization

Abstract

Recently, channel state information (CSI) has been adopted as an enhanced wireless channel measurement instead of received signal strength (RSS) for indoor WiFi positioning systems. However, although CSI contains richer location information, a challenging problem is the severe dynamic range and fluctuation among the high-dimensional channels, which may degrade accuracy and cause overfitting problems. This paper proposes a novel algorithm for improved fingerprinting-based indoor localization. The proposed algorithm decomposes the CSI sequence using the multilevel discrete wavelet transform (MDWT) and normalizes the wavelet coefficients by employing histogram equalization. The robust features were then extracted by reconstructing CSI through the inverse MDWT of the normalized coefficients. We demonstrate the effectiveness of the proposed algorithm through experiments. The results show that the proposed algorithm outperforms traditional RSS, CSI, and two CSI-based algorithms, FIFS and MIMO.