| Summary: The multipath reflections in visible light communication pose a significant problem for RSS-based indoor positioning systems. In this thesis, effects of the multipath reflections on indoor positioning systems have been tried to be eliminated. For this purpose, a differentiator filter-assisted RSS-based indoor positioning system and an artificial neural network-assisted indoor positioning system have been proposed.
Thanks to the differentiator filter-assisted RSS-based indoor positioning system, the distortions of the multipath reflections on the optical signal have been eliminated, so the positioning errors have been decreased and reduced to the millimetric order. At the same time, the positioning estimation performed by the artificial neural network assisted indoor positioning system has been found to demonstrate a great performance as with the former system. However, this method requires channel information along with previously tested input-output maps for different environments.
In this thesis, an indoor positioning system scenario was considered and all simulation studies were performed according to this scenario. In the second part of the thesis, the modeling stages of a wireless optical channel have been introduced and the system designs of the proposed positioning techniques have been realized. Last but not least, the proposed systems have been examined, evaluated and compared with each other in terms of the positioning errors. |
