Indoor Localization Algorithm Based on Information Gain Ratio and Affinity Propagation Clustering.

Area of Science:

• Computer Science
• Electrical Engineering
• Signal Processing

Background:

• Indoor positioning systems commonly use Wi-Fi access points (APs) for fingerprinting.
• Redundant APs increase computational load and offer diminishing accuracy improvements.
• Existing methods lack efficient strategies for AP selection in fingerprint databases.

Purpose of the Study:

• To propose a novel method for eliminating redundant APs in indoor positioning systems.
• To reduce the computational load and database size without compromising positioning accuracy.
• To enhance the accuracy and reduce the error of indoor positioning estimation.

Main Methods:

• Calculating AP location discrimination capability and combining it with coverage for AP elimination.
• Utilizing affinity propagation clustering for coarse positioning.
• Employing Bayesian methods for fine positioning.

Main Results:

• The AP removal strategy reduced fingerprint database size by up to 72.78% while maintaining positioning accuracy within a 5% fluctuation.
• The proposed positioning method reduced average positioning error by 11% to 39% compared to GMM, K-Means, and pointwise algorithms.
• Experimental validation on real-world and benchmark datasets confirmed the effectiveness of the proposed methods.

Conclusions:

• The developed AP elimination strategy is effective in optimizing fingerprint databases for indoor positioning.
• The combined coarse and fine positioning approach significantly improves accuracy and reduces errors.
• This research offers a practical solution for efficient and accurate indoor positioning.