A case study of indoor positioning in an unmodified factory environment

This study investigates the performance of Wi-Fi-based fingerprinting location estimation in an unmodified factory environments of Hon-Hai Precision Industry. The positioning system contains four components: temporal filter, dynamic clustering, access point selection, and pattern-matching. We evaluated the performance using realistic Wi-Fi power measurements. Results show that, although reasonable accuracy is achieved (around 72-percentil error is smaller than 7 meter), there always exist much large errors (e.g., over 13 meter). We argue that for a large factory environment, sufficient number of the-same-floor APs is required for many indoor location-based applications such as labor tracking.