Study On RSS-Based Indoor Location Technology In Large Buildings

There have been increasing innovations of Internet application because of the rapid growth of mobile Internet focusing on smart phoncs terminals and open applications. Booming LBS (Location Based Service) application has become one of the most popular applications in mobile Internet field. As a huge number of large buildings appear nowadays, LBS is not only limited in outdoor areas. However, traditional satellite positioning systems always fail in indoor positining. The study of RSS (Received Signal Strength) based wireless location technology becomes very important and practical due to the large demand for indoor LBS and the more sophisticated WLAN (Wireless Local Area Neiworks) technology.Based on the frontier progress in information field, a series of algorithms have been proposed for LBS in large buildings in this dissertation, including construction method for location fingerprint database, location fingerprint matching algorithm in the complex indoor environment and multi-sensor indoor tracking problem. The main contributions are as follows:1) Construction method for location fingerprint databaseA weighted mean-shift algorithm for location fingerprint construction is proposed in this dissertation in order to reduce the collection workload and get a relatively complete location fingerprint data in a shorte time. Meanwhile, an attribute reduction method for location fingerprints based on neighborhood rough set theory is presented in the dissertation in order to reduce location fingerprint storage for mobile handheld terminals as well as computational complexity. The experiments show the superiority in accuracy of our method compared with original AP selection strategy.2) Location fingerprint matching algorithm based on LRSMLA new 11-graph regularized semi-supervised manifold learning (LRSML) method for location fingerprint matching is proposed in this dissertation for noise corruption problem of received signal strength (RSS) in complex environment of large buildings. The construction process of 11-graph is assumed to be unsupervised without harnessing any data label information and uncovers the underlying sparse relationship of each data. Utilizing both labeled and unlabeled information improve the localization accuracy and generalization capability, so the LRSML method has the potential to convey more discriminative information compared to conventional methods. Experimental results show the superiority of our method on robustness, positioning performance and generalization ability over several current state-of-art methods.3) Strong indoor tracking method based on PDRA three-dimensional discrimination method based on built-in three-axis accelerometer and electronic compass of smart phone is proposed in this dissertation to improve positioning autonomy. Location information can still be obtained using our method even under the condition that wireless signal is weak at the border region or access point does not work because of losing power. A federal Kalman method based on the pedestrian sensor information is further presented by integrating the pedestrian dead reckoning and RSS-based location fingerprint positioning technology. Experimental results show the higher tracking precision and better user experience of our method compared with original Kalman filter.An indoor location-aware system for large buildings is developed in order to verify the location theory and method presented in this dissertation, consisting of server, PC clients, mobile handheld clients and location fingerprint database. The dissertation details the system architecture, configuration and functions, as well as the achievements of system software and each functional module. The description of the experimental scene and system performance is also given in detail.Finally, the research work is summarized and the direction for further research is pointed out.