Indoor location Wi-Fi fingerprinting using invariant received signal strength

Abstract

The use of Wi-Fi Received Signal Strength (RSS) for location fingerprinting in GPS-denied indoor environments has drawn much attention recently as an enabler of various location based personal services with handheld/wearable communication devices. However, the instability of RSS incurred by highly mutable channel characteristics, due to random spatiotemporal disturbances, hampers a wide-spread adoption of RSS based location fingerprinting to real world applications. To cope with the RSS instability, this study proposes a new approach based on the concept of "the invariant RSS statistics". The invariant RSS statistics are defined at individual calibration locations as the statistics of the particular RSS readings collected with minimal of random spatiotemporal disturbances. The invariant RSS statistics consequently formulated as the invariant reference class patterns at individual calibration locations for localization. The experimental results show that in comparison to conventional approach, the proposed approach performs better by 17% in localization success rate, and the performance does not deteriorate by time varying.