A Novel Non-Contact Multi-User Online Indoor Positioning Strategy Based on Channel State Information.

Area of Science:

• Wireless communication and networking
• Machine learning applications
• Indoor positioning systems

Background:

• IEEE 802.11bf WiFi is emerging for indoor positioning.
• Multi-user online positioning in real wireless environments presents challenges.
• Accurate indoor localization is crucial for various applications.

Purpose of the Study:

• To propose an optimized indoor positioning sensing strategy.
• To introduce a novel machine learning (ML) method, NKCK, for enhanced positioning accuracy.
• To evaluate the performance of the NKCK method against traditional approaches.

Main Methods:

• The NKCK method integrates Neighborhood Component Analysis (NCA) for dimensionality reduction, K-means clustering, and K-nearest neighbor (KNN) classification with cross-validation (CV).
• Channel State Information (CSI) amplitude fingerprinting is utilized.
• The strategy includes an optimized preprocessing step.

Main Results:

• The NKCK method demonstrated significant error rate reductions compared to various traditional methods.
• Achieved error rate reductions include 82.4% vs. Naive Bayes, 85.0% vs. Random Forest, and 72.1% vs. Support Vector Machine.
• Outperformed PCA-based and LDA-based CSI amplitude fingerprinting methods.

Conclusions:

• The proposed NKCK method offers superior performance for indoor positioning compared to existing techniques.
• The system shows promise for multi-user online positioning in complex wireless environments.
• Further research is needed to address challenges in detecting dynamic human activities like tracking due to CSI sensitivity.