Adaptive Fitting Reference Frame for 2-D Indoor Localization Based on RFID

Abstract

Abstract-Indoor localization information is important to many different industries. Ever-rising demand from both custom and supplier pose great challenge for indoor localization technology. The emergence of rapid development of Radio Frequency Identification (RFID) technology enriches indoor positioning methodology. Traditional indoor localization algorithm mainly based on the signal propagation property. Among all of positioning algorithm, LANDMARC is most promising one but still has some limitation such as multipath effect, noise form environment and quantity constrain of tags. This paper identifies previous RFID-based indoor localization algorithm and proposes a new algorithm aiming at delivering above limitation to estimate objective position indoor environment. It use RSS propagation curve to obtain RSS value of virtual reference tag, and then estimate target tags position. To evaluate its performance, a comparison between LANDMARC and proposed algorithm was conducted. Experiment result show proposed algorithm is more accurate than LANDMARC.

RFID-enabled indoor positioning method for a real-time manufacturing execution system using OS-ELM

“…The positioning systems to be compared include ELM based method [20], the LANDMARC algorithm [12], the Adaptive Fitting Model (AFM) [13], RADAR [35], Support Vector Regression (SVR) [36], and Backpropagation (BP) Neural Networks [37]. All evaluation processes were conducted in a Matlab environment running in a Windows 7 machine with a 3.00 GHz CPU and 4 GB RAM.…”

“…Recently, the Adaptive Fitting Model (AFM) was proposed that adopts a path loss model and describes the virtual reference tags as an RSS propagation curve rather than adopting empirically pre-set interval ratios [13]. Signal collection, outlier elimination, and curve fitting are the three constructive processes for the AFM model.…”

“…Traditional RFID positioning systems employ the idea of using reference tags to provide neighbourhood reference positions (such as the landmarks used for navigation in our daily lives), and compare relative RSS to compute the position of the target object [12,13]. For each reader, the RSS differences from the target object, SMO t , to every reference tag, Rt j (j=1,2,…,n, j≠t), need to be computed and ranked to find the neighbourhood reference tags for further positional mapping.…”