M-Show: A system for accurate position estimation in multi-floor buildings

This paper presents M-Show, a Wireless LAN based system for locating users inside multi-floor buildings. It makes improvements in the sampling, processing and storing of empirical signals to previous studies, and builds a Fuzzy Empirical Signal Model that gives more credible description of signal distribution. By dividing the map into different sized Intelligent Areas (iAreas), M-Show is able to achieve alterable positioning granularity within the same map. It also unprecedented adopts AP Close Range Pre-Locating strategy to quickly locate the users inside a certain small scope around an AP. In addition, since statistical calculation methods are inefficient in distinguishing vertically overlapping positions on different floors, which is not tackled in previous locating systems, we introduce the Fluoroscopy to rectify the locating result of probability calculation. M-Show is deployed in Shanghai Science and Technology Museum, and experimental results show that with reduced average position query time and lower computation cost, M-Show achieves a high locating accuracy of 93% probability within 3 feet around the user's actual position.