Skip to Main Content
User localization information is an important data source for ubiquitous assistance in smart environments. This paper proposes a device-free passive user localization approach based on room-equipped passive RFID instead of battery powered hardware. Based on this approach recent work tried to formulate physical model based localization algorithms. These approaches suffer from their inability of integrating environmental changes like the deployment under moving experimental conditions. On the other hand most model based approaches have a certain trade-off between a high localization precision and computational complexity. In this work we try to formulate a training based approach to the problems with the help of artificial neural networks. Special representatives like multi-layered perceptrons are applied to a wide range of problems where it is difficult to model the underlying physical condition completely. We present a perceptron implementation for the purpose of user localization and conduct first results with different model parameters and functions.