Angel Ray Projection Based Crowdsourcing Collaboration For Indoor WLAN Positioning Algorithm

Wireless Local Area Network(WLAN) technology has been in a state of rapid development in the field of wireless communications since its inception. The rapid popularization of intelligent terminal and mobile Internet, and the extensive layout of indoor WLAN access point(AP) have brought great convenience for the study of indoor WLAN positioning technology. Furthermore, users do not need to add additional equipment, which can be able to meet the positioning and navigation needs in a lower cost. At present, it is widely used in the indoor WLAN location algorithm, but there are two main problems in the fingerprint location algorithm: first, the establishment of off-line fingerprint database is time-consuming and laborious; second, the complexity of the indoor signal propagation in the on-line stage, which leads to the low accuracy of location matching. This thesis is aiming at these two problems. Based on the fingerprint location algorithm, a new algorithm of angel ray projection based crowdsourcing collaboration for indoor WLAN positioning is proposed, to achieve the purpose of high-precision positioning of the reference points in the sparse conditions.The proposed algorithm consists of off-line phase and on-line phase. The off-line phase consists of two steps: The construction of the fingerprint database and the preprocessing of the fingerprint database. In the stage of construction of fingerprint database, this thesis presents an indoor WLAN fingerprint database construction based on virtual AP. In order to extend the sparse fingerprint database into a multi direction mixed fingerprint database, linear regression analysis is conducted to obtain the total minimum mean square error of the best virtual AP position through the sparse fingerprint database in off-line phase. And then use the multi-direction signal propagation model to construct the multi-direction fingerprint database for each virtual AP. This method reduces the cost of building database in the offline stage under the precondition of ensuring accuracy. In the preprocessing stage of fingerprint database, the affinity propagation clustering based on the similarity of Tanimoto is used to preprocess the multi-direction hybrid fingerprint database, which forms the clustering center, reduces the matching time of the on-line phase, and improves the real-time performance of the algorithm.The on-line phase consists of three steps: first, the point of crowdsourcing collaboration uses the clustering center formed by the off-line phase to carry out the rough location and the K-Nearest Neighbor location algorithm to get its own estimation position for precise positioning; second, by means of the calculation of the magnetic force meter of the mobile terminal Micro Electro Mechanical Systems, we can obtain the angle information of the collaboration point with respect to the point to be located. third, constructing the model of angel ray projection based crowdsourcing collaboration for indoor WLAN to get the final location of point to be located. In this thesis, we combine the angle information and the received signal strength information to achieve the goal of high precision positioning. And verifying in the Chongqing University of Posts and Telecommunications Yi Fu Building 5 floor, this algorithm is within 3 meters of cumulative probability of error reaches 89.54%, the average positioning error reaches 1.69 meter, which verifies the validity of the algorithm.