Location-Based Service (LBS) has become one of big issues in this year. By the worldwide proliferation of smart phones, many location-based services are spread out all over the platforms. They support several services using location information estimated from various sensors on smart phone. Naturally, location estimation is the most essential to the success of LBSs. WLAN (IEEE 802.11) based Positioning Systems (WPS) are the most suitable technology for LBSs, compared with other technologies. WLAN-based positioning systems have many advantages in terms of coverage and costs. And WPS can estimate location within small error distance without distinction between indoor and outdoor. Most of researches on WLAN based positioning system are focus on proposing new classifier to improve positioning accuracy or effective fingerprint data generation. WPSs which are proposed in previous work didn’t consider some constraints of mobile devices which are actually used and some environmental conditions in WPSs. Mobile devices like smart phone don’t have enough memory for carrying all fingerprints data in large-scale environment, such as department store, huge shopping mall. And, when we develop real-time localization system using existing WPS, it is hard to guarantee acceptable response time on mobile devices which have low computation power. Thus, we have to consider two constrains, resource consumption and response time, to realize real-time localization system. In this thesis, we propose two fingerprint indexing methods for large-scale radiomap. Our methods effectively reduce search space and computation time compared with clustering methods. As the result applied to COEX, our methods are 6 times faster than others with similar accuracy of non-indexing case which is searching all fingerprints.