Underground Coal Mine Fingerprint Positioning Based on the MA-VAP Method.

Area of Science:

• Wireless communication networks
• Indoor positioning systems
• Signal processing

Background:

• Coal mine WLANs suffer from sparse access points (APs), hindering Wi-Fi fingerprint positioning.
• Existing Virtual Access Point (VAP) methods generate Received Signal Strength (RSS) values based on single AP mapping, ignoring signal correlations.
• This limitation leads to incomplete fingerprint modeling and real-time RSS collection, impacting positioning accuracy.

Purpose of the Study:

• To propose a novel Multi-Association Virtual Access Point (MA-VAP) method for improved Wi-Fi positioning in environments with sparse AP deployment.
• To address the limitations of the VAP method by incorporating multi-association signal correlations.
• To analyze the impact of VAP quantity and arrangement on positioning accuracy.

Main Methods:

• Developed the MA-VAP method, calculating a multi-association coefficient based on RSS correlations between VAPs and multiple APs.
• Utilized a multi-association function to generate VAP RSS values, using real-time multi-AP RSS as input.
• Evaluated positioning accuracy using Weight K-Nearest Neighbors (WKNN) and Kernel Principal Component Analysis (KPCA) algorithms.

Main Results:

• The MA-VAP method demonstrated superior positioning performance compared to the VAP method for identical VAP arrangements.
• In non-line-of-sight (NLOS) environments, MA-VAP achieved 90% positioning accuracy within 4.5m (WKNN) and 3.5m (KPCA).
• Positioning accuracy improved by 10% (WKNN) and 22.2% (KPCA) compared to the VAP method.

Conclusions:

• The MA-VAP method effectively resolves Wi-Fi positioning challenges in areas with sparse AP deployment.
• The proposed method significantly enhances positioning accuracy by considering multi-association signal characteristics.
• MA-VAP offers a cost-effective solution for improving indoor positioning in challenging environments like coal mines.