Performance Modeling and Design of Backscatter RFID Systems:A Statistical Approach

Abstract

The current models for specifying and estimating performance of RFID systems are based purely on EM theory (e.g., using Friis free space equations) or statistical experimental modeling principles. Models based on EM theory are limited to specifying power received at the tag under certain simple, idealized conditions, and do not provide estimates of read-rates, which are de-facto industry quantifiers of an RFID system performance. The estimates of power levels do not consider uncertainties inherent to an industrial RFID system. On the other hand, the statistical models, being purely data-driven, suffer from lack of generalizability as their results cannot be extrapolated. This thesis investigates a statistical approach for assessing the system performance and proposes an analytical probabilistic model based on Friis free space expression that captures the uncertainties existing in gain of the tag antenna, power of the reader antenna, frequency hopping, etc. Finally the thesis suggests a set of techniques to increase read-rate probabilities of RFID tags when placed on metal objects or in the presence of highly metallic environments.