Research On The Key Techniques Of WLAN Indoor Localization Based On Location Fingerprint

With the development of mobile Internet technology, there is an increasingly demand for location-based services(LBS). LBS are widely used in various occasions such as social software, medical aid, market guide and warehouse management. Global navigation satellite system is mainly used in outdoor environment. However, due to the satellite signal is blocked by buildings, the satellite positioning system cannot cover the indoor environment. Owing to the rapid development of Wireless Local Access Network(WLAN), wireless access points(APs) have been widespread arranged in indoor environment. Compared with other indoor positioning system, location fingerprint-based WLAN indoor wireless positioning system has the advantages of low hardware costs, wide distribution, and high accuracy.In this thesis, we take WLAN and location fingerprinting positioning technology as an object of the research. According to the deficiencies of the current technology, combining the theory and methods of pattern recognition and machine learning, we propose the corresponding solution. The main work includes three aspects.For the uncertainty of APs’ received signal strength(RSS), an APs selection method named Stable MaxMean is proposed. Stable MaxMean APs selection algorithm is used to select the top m APs which have highest mean RSS and stable. Experimental results show the APs selection method achieves significant accuracy, stability improvement, and time saving accomplishment.For over-fitting and low generalization ability problems because of the existence of singular points, a hybrid clustering algorithm K Nearest Neighbor-Support Vector Machine,(KNN-SVM) based on fingerprint database is proposed. Firstly, removing singular points using KNN method. Secondly, compute the user’s location based on SVM algorithm. Experimental results show that KNN-SVM algorithm achieves significant positioning accuracy, while reducing positioning error.On the basis of the above research, an indoor positioning system based on the Android platform named IOTLocation is designed. Experimental results show that the system has high positioning accuracy, good scalability and low complexity.