In Wi-Fi fingerprinting, a set of signal strengths of access points, termed RSSI (Received Signal Strength Information), is measured at a target for indoor location estimation. Traditionally, the RSSI is compared in aggregate as a vector with the ones of the reference points (for closest neighbor search). As signal noise in site survey is inevitable, this often leads to a disperse set of closely matched reference points, resulting in unsatisfactory estimation accuracy.

Observing that the target measures the RSSI at the intersection of AP signal contours, we propose in this work a highly accuate localization scheme termed Contjunction. By considering measurement noise in site survey, Contjunction uses a novel closeness measure in signal space. It efficiently identifies a tight...[ Read more ]

In Wi-Fi fingerprinting, a set of signal strengths of access points, termed RSSI (Received Signal Strength Information), is measured at a target for indoor location estimation. Traditionally, the RSSI is compared in aggregate as a vector with the ones of the reference points (for closest neighbor search). As signal noise in site survey is inevitable, this often leads to a disperse set of closely matched reference points, resulting in unsatisfactory estimation accuracy.

Observing that the target measures the RSSI at the intersection of AP signal contours, we propose in this work a highly accuate localization scheme termed Contjunction. By considering measurement noise in site survey, Contjunction uses a novel closeness measure in signal space. It efficiently identifies a tight region which best represents the target RSSI by computing the junction of its individual signal strengths (instead of in aggregate) without leading to a disperse set of reference points. Contjunction is computationally efficient, and has simple closed-form solutions for a wide variety of noise distribution such as uniform, normal or log-normal. Extensive simulation and large-scale experiments in both campus and airpport environment show that Contjuntion significantly outperforms other recent schemes substantially.