Signal Propagation Model Calibration under Metal Noise Factor and Advanced Positioning Algorithm of RFID System with Metal Noise Cancellation By Ni Xue M-B2-5548-2 Thesis Supervisor: Professor Wong SengFat M.Sc. in Electromechanical Engineering Radio frequency identification (RFID) technology as an effective indoor localization solution is acquiring increasing attention for its low cost and compactness. It has become a significant technological tool for enterprise to raise the management level of logistics and supply chain, reduce cost and strengthen their competitiveness. However, most of those industrial applications are used in metallic environment which will degrade the performance of RF-based localization, because the most common location algorithm methods, like LANDMARC and VIRE, do not work well under metal noise factor. Unlike earlier attempts to resist the effect of metal by absorbing and isolating material at high cost, this work proposes a new improved algorithm suitable for operation on metallic surfaces. Firstly, a new calibrated RSSI propagation curve under metal noise factor has been obtained from practical experiments. By taking advantage of this metallic effect characteristic curve, RSSI value for calculate weight are effectively converted because the weighted distance is more accurate for calculating coordinate of tracking tags. Secondly, cubic spline interpolation has been used instead of linear interpolation to obtain Received Signal Strength Indication (RSSI) of virtual reference tags. Finally, without limited by a fixed threshold, a M-tupling smallest Ek as threshold to select possible regions which have the smallest difference of RSSI value between their region and tracking tag for calculating weighting factor. In the way, the imprecise location accuracy is avoided which is caused by filtering too many or too few possible region in metal environment. As a result, the new approach enhances the precision of indoor location under metal noise factor by 90％ over original method in simulation and 26％ in practical experiments.