AG Kommunikationstheorie


Tracking Methods for a Software Defined Radio Based Localization System


The localization of a radio transmitter in a sensor network is a difficult task because the speed of the transmitted electromagnetic waves yields large position errors from small mistakes in timing measurements. Since the sampling clocks of the receivers cannot be perfectly synchronized, methods to reduce the error must be found. This thesis develops a Kalman filter based tracking approach to reduce the noise of the location estimates for dynamic transmitters in a Software Defined Radio (SDR) based real-time system for Time Difference of Arrival (TDOA) localization. It begins with introducing basic methods for localization and tracking. Consequently, the specific problems of TDOA localization algorithms and practical issues of the deployed system are discussed. A mayor challenge in the localization is the dependency of the error on the network topography and the current location of the object. This is modeled using the dilution of precision concept. In addition, large outliers occur under real-world conditions that can severely harm the performance of the tracking. Therefore, pre-filtering techniques to suppress such errors prior to the tracking stage are shown. Furthermore, the interplay of all components is explained and all parameters are tuned to ensure good performance. Simulations for verifying the ideas are presented and their results are analyzed. For this a movement model using spline interpolation has been developed. Finally, the real-time localization system is described with focus on the experimental setup and the modifications that have been introduced for the tracking. Results from outdoor localization experiments are presented and discussed.

zurück zur Terminübersicht