Movement Path Data Generation from Wi-Fi Fingerprints for Recurrent Neural Networks

Area of Science:

• Computer Science
• Machine Learning
• Signal Processing

Background:

• Indoor positioning systems often suffer from significant errors.
• Recurrent Neural Networks (RNNs) can improve accuracy by learning from sequential data.
• Traditional Wi-Fi fingerprinting lacks the sequential data required for RNNs.

Purpose of the Study:

• To develop a method for generating sequential data from Wi-Fi fingerprints for RNN-based indoor positioning.
• To reduce positioning errors in indoor environments.

Main Methods:

• Proposed a movement path data generation technique for Wi-Fi fingerprint data.
• Utilized K-means clustering to divide the indoor environment.
• Created a cluster transition matrix to manage computational complexity.

Main Results:

• Successfully generated sequential data suitable for RNN models from Wi-Fi fingerprints.
• Demonstrated a method to overcome the limitations of standard Wi-Fi fingerprinting for sequential analysis.
• Addressed computational challenges in creating adjacency matrices for large datasets.

Conclusions:

• The proposed data generation technique enables effective use of RNNs for Wi-Fi fingerprint-based indoor positioning.
• K-means clustering efficiently handles large datasets and reduces computational load.
• This approach significantly reduces indoor positioning error.