In order to build intelligent spaces, it is necessary to know where the distributed sensors are placed in the space. Also, the positions of sensors must be known in a unified world coordinate system. For this reason, we have the system is based on map sharing and matching between a mobile robot and distributed sensors. However the system is subject to effects of the map with distortion and errors. In this paper, we have introduced a system for easily estimating locations of sensors for constructing large-scale intelligent spaces. Especially, position estimation improvement based on more accurate global maps is introduced.