CSI-based sliding window fingerprinting method tailored for a signal blocking environment in VLP systems

Area of Science:

• Electrical Engineering
• Computer Science
• Signal Processing

Background:

• Visible light indoor positioning systems often suffer from reduced accuracy due to signal occlusion.
• Received signal strength (RSS)-based fingerprinting algorithms are particularly vulnerable to line-of-sight (LOS) signal blockage.
• Robust localization methods are needed for reliable indoor navigation.

Purpose of the Study:

• To enhance the accuracy and robustness of indoor positioning in visible light systems.
• To address the performance degradation caused by line-of-sight (LOS) signal occlusion.
• To propose a novel algorithm that leverages channel state information (CSI) for improved localization.

Main Methods:

• A sliding window fingerprinting (SWF) algorithm is developed, combining CSI with sliding matching techniques.
• The SWF algorithm uses a sliding window to match received CSI with a database, reducing interference from LOS signal loss.
• A weighted sliding window fingerprinting (W-SWF) variant is introduced to account for varying path contributions in CSI.

Main Results:

• The proposed SWF and W-SWF algorithms demonstrate superior performance in a simulated indoor multipath environment.
• Mean errors of 0.20 cm and 1.43 cm were achieved even when 1 or 2 LED LOS signals were blocked.
• SWF algorithm showed over 90% improvement in mean error and root mean square error (RMSE) compared to traditional RSS, WKNN, and ARWKNN algorithms.

Conclusions:

• The SWF algorithm effectively enhances indoor positioning accuracy and robustness in visible light systems.
• Leveraging CSI and sliding window matching provides a significant improvement over existing methods, especially under signal occlusion.
• The proposed W-SWF further refines accuracy by weighting CSI path contributions, offering a promising solution for reliable indoor localization.