Experimental Study on RSS Based Indoor Positioning Algorithms

This work compares the performance of indoor positioning systems suitable for low power wireless sensor networks. The research goal is to study positioning techniques that are compatible with real-time positioning in wireless sensor networks, having low-power and low complexity as requirements. Map matching, approximate positioning (weighted centroid) and exact positioning algorithms (least squares) were tested and compared in a small predefined indoor environment. We found that, for our test scenario, weighted centroid algorithms provide better results than map matching. Least squares proved to be completely unreliable when using distances obtained by the one-slope propagation model. Major improvements in the positioning error were found when body influence was removed from the test scenario. The results show that the positioning error can be improved if the body effect in received signal strength is accounted for in the algorithms.