Indoor Localization by UHF RFID Technology: An approach from probability and statistics

Abstract

Radio Frequency Identification (RFID) is a rapidly growing market for many applications. Like bar codes, it enables automatic identification of physical objects, but it does not require a line of sight, and it has a much bigger range. This bigger read range makes it suitable for localization purposes. In our research we present an efficient localization procedure for indoor environments, based on signal strength measurements of the radio communication. The received signal strength is very noisy, an exponential probability distribution appears to model it well. Hence, the localization can be realized by applying the method of maximum likelihood to signal strength measurements at multiple RFID readers. Maximizing the likelihood is an optimization problem which we solve numerically by a quasi-Newton method. Furthermore, we show by simulated signal strength measurements the dependence of the localization accuracy on the physical parameters. Finally, we check the results by solving the localization problem by an alternative method, neural networks, which is a black box solver. The research was done at Intellifi company, which provided a modern RFID system.