Pedestrian Positioning Using a Double-Stacked Particle Filter in Indoor Wireless Networks

Area of Science:

• Computer Science
• Electrical Engineering
• Robotics

Background:

• Indoor pedestrian positioning is challenging due to sensor errors from microelectromechanical systems (MEMS) and radio-frequency (RF) signal variations.
• Existing methods often combine received signal strength (RSS) fingerprinting with dead reckoning (DR) using Bayes filters, but particle filters (PF) can be computationally intensive.

Purpose of the Study:

• To develop a more computationally efficient and accurate indoor pedestrian localization scheme for mobile phones.
• To address the limitations of traditional methods, including MEMS sensor bias and RF signal degradation.

Main Methods:

• The proposed system integrates RSS fingerprinting and DR with a novel Double-Stacked Particle Filter (DSPF).
• DSPF estimates user position by fusing noisy data from RSS and DR, utilizing proposal and target distributions for improved accuracy.
• The algorithm is implemented on a mobile phone platform for practical indoor navigation.

Main Results:

• The DSPF algorithm demonstrated superior localization accuracy compared to Kalman filtering-based methods.
• It achieved competitive accuracy with standard PFs but with significantly higher computational efficiency.
• Experimental results confirmed the DSPF's ability to provide reliable and accurate indoor positioning.

Conclusions:

• The Double-Stacked Particle Filter (DSPF) offers an effective solution for accurate and computationally efficient indoor pedestrian localization.
• This method enhances mobile phone-based navigation by overcoming the limitations of existing techniques.
• The DSPF presents a promising advancement for real-world indoor positioning applications.