In this paper, we propose a new robust algorithm to improve the localization accuracy without using additional materials. This algorithm is based on a probabilistic approach which only uses the read count (number of detection) of reference and target tags. The localization is performed through two steps. The first step consists of estimating a local position of object tags in a given zone, while in the second step, all estimated local positions are used to calculate one global position. Furthermore, a complementary study concerning the influence of reference tags positions, the movement nature and the coverage of the work area is presented and discussed. For the proof of concept, a series of tests were elaborated with the proposed localization system created comprising conventional passive EPCglobal Class-1 Gen-2 UHF RFID tags, a commercial UHF RFID reader and a laptop. Measurements are carried out in an indoor environment where the 5.5 m×1.25 m measurement zone was placed on drywalls. The experimental results show that the present algorithm gives higher localization accuracy than the conventional ones.
Read full abstract