Pre-processing-for-AngLoc-Indoor-Location-Fingerprinting-using-Wi-Fi-CSI

AngLoc, an AOA-aware probabilistic indoor localization system using Wi-Fi CSI, can predict the indoor location. Before we construct the offline radio map, it is important to do CSI pre-processing, the techniques which serve as the precondition to attain superior localization performance.

Data Setup

• end-to-end MIMO-OFDM wireless transceiver for IEEE 802.11 n/ac
• 56 subcarriers
• CFR (channel frequency response) of CSI: $$H(f)=|H(f)|e^{j\angle H(f)}$$
• CIF (channel implies response) of CSI: $$h(\tau)=\Sigma\alpha_ie^{-j\varphi_{i}\delta(\tau-\tau_i)}$$
• 4 reference points
• 2 TX antennas, and 2 RX antennas
• one position collect 3000 CSI packets

System Design

Noise Removal

• tap filtering

Phase Calibration

• SFO removal
• STO removal
• CFO removal

Result

We show the CFR and CIR figure, and there are four step figure: original, after tap filtering, after SFO removal and after STO removal.
https://github.com/mandy1223ks/Pre-processing-for-AngLoc-Indoor-Location-Fingerprinting-using-Wi-Fi-CSI/blob/main/presentation.pdf