Smartphone-Based Indoor Localization within a 13th Century Historic Building

Area of Science:

• Computer Science
• Electrical Engineering
• Geographic Information Systems

Background:

• Indoor localization systems are crucial for navigation in GPS-denied environments.
• Existing methods like fingerprinting can be time-consuming and complex to set up.
• Accurate and efficient indoor positioning remains a significant challenge.

Purpose of the Study:

• To present an updated indoor localization system for smartphones.
• To improve positioning accuracy and reduce setup time.
• To address challenges like multimodal densities and sample impoverishment in particle filters.

Main Methods:

• Utilized a particle filter for recursive state estimation with probabilistic sensor models.
• Implemented an approximation scheme for kernel density estimation for exact position estimation.
• Developed a novel state transition model based on navigation meshes and activity recognition (barometer, accelerometer).
• Employed an optimization scheme with reference measurements for access point signal strength prediction.

Main Results:

• Achieved small positioning errors even in challenging, historic building environments.
• Demonstrated a setup time of less than 120 minutes for 2500 m² walkable area.
• Successfully tackled issues like multimodal densities and sample impoverishment.
• Validated the system through 28 measurement series across four test walks.

Conclusions:

• The updated indoor localization system offers a fast-deployable solution with reasonable accuracy.
• The novel navigation mesh and activity recognition enable continuous and smooth floor changes.
• The system provides a fail-safe solution for indoor positioning in diverse conditions.