Indoor Localization with Passive Sensors.

摘要

In this thesis, a framework is described that is designed to perform indoor localization in the SmartCondo(TM). A significant aspect of the framework is that it mainly operates on the basis of binary sensors -- including motion sensors and occupancy sensors -- and it primarily involves geometric computations. In addition, switch-type sensors have been incorporated. We have specifically designed and implemented a geometry library to facilitate the necessary computations, as well as a simulation tool to simulate the environment and its sensors. Compared to previous related research work, we adopt a more realistic environment model, as well as models for a person's body, and models for the sensors. In the experiments conducted, when the sensors are assumed to behave in an ideal fashion, we have achieved 67cm and 49cm as mean localization error for minimum coverage and dense coverage sensor configurations respectively. Under a more realistic sensor behavior model the corresponding numbers are 69cm and 62cm respectively.