Device-Free Indoor Localization of CSI Based on Limited Penetrable Horizontal Visibility Graph

Abstract

Most of the current indoor localization methods based on channel state information (CSI) utilize the amplitude and phase of each subcarrier individually as location fingerprints, but the correlation between adjacent subcarriers also contains essential location information. Therefore, this paper proposes a device-free indoor localization method based on CSI and limited penetrable horizontal visibility graph (LPHVG), which does not require the subject to carry a device and can be implemented with only a single access point (AP). Firstly, we model the frequency correlation between subcarriers by LPHVG algorithm and construct a CSI-based complex network. Secondly, the topology of the complex network is utilized to analyze the relationship between CSI adjacent subcarriers and extract network features, which are combined with statistical features as fingerprint information to characterize different locations. Finally, the method is combined with support vector regression (SVR) to realize indoor localization in different environments. The experimental results show that the proposed method can significantly improve the localization accuracy, and has an excellent performance in different indoor scenarios.

MFFALoc: CSI-Based Multifeatures Fusion Adaptive Device-Free Passive Indoor Fingerprinting Localization

“…As previously mentioned, CSI-based Device-free Passive Fingerprinting Localization technology stands as the most prevalent Wi-Fi positioning method, representing a prominent trend in the field, with many researchers carrying out research on it. Liu et al [16] introduced a device-free indoor localization method using CSI and a visibility map algorithm with finite penetrability levels. Their approach involved modeling frequency correlation between subcarriers, constructing a complex network based on CSI, and using network topology to analyze relationships among subcarriers.…”

“…In recent times, CSIbased device-free passive localization has garnered prominence as one of the most practical and convenient approaches for addressing indoor personnel localization and sensing challenges. It mainly can be classified into two main categories: radio signal propagation model-based solutions [13]- [15], and fingerprint-based solutions [16]- [22]. Radio signal modelbased solutions employ statistical models to estimate a user's location using CSI measurements, eliminating the need for fingerprint databases and reducing site surveys.…”