Improved Fingerprint Localization by Using Static and Dynamic Segmentation

Abstract

Fingerprint based localization approach is improved by using segmentation across the sensing area. The approach requires a radio map prepared by the RSSI values at fingerprint points. Localization algorithms are applied with these RSSI values at object and fingerprint points. Sensing area is divided into sub areas defined as static segments. A feature function is identified for each segment as the range of RSSI values received at its fingerprint points. The localization accuracies achieved with the static segmentation is 1.2 meter while the accuracies achieved with the classical fingerprint technique is 1.8meter. Dynamic segmentation is generated automatically by choosing similar RSSI values at fingerprint points across the sensing area. Close RSSI values are gathered together dynamically to utilize segments where the object search is carried out. The localization accuracies achieved with dynamic segmentation is 0.9 meter.