Application Of Fingerprint Positioning System Based On Wifi In Indoor Location

With the rapid development of mobile communication technology, LBS(Location Based Services) based on wireless sensor network has become more and more popular. Localization technology is the foundation for LBS, so how to achieve precise positioning is the key to LBS. GPS can provide enough accurate location services in outdoor environment, but in valley downtown or inside the building GPS satellite signal is difficult to be received, so it is almost impossible to use GPS for positioning in those areas. In order to reduce the blind area of GPS and provide high-precision positioning services in indoor environment, this paper firstly studies classic Wifi fingerprint location algorithm, and then designs a new fusion location algorithm based on KNN and Bayes to solve several major problems that exist in the original algorithm. The experiments have proved that the new algorithm not only can meet the need of location accuracy, but also can reduce the complexity of computation. The main work is as follows:(1) In indoor environment, the propagation of Wifi signal will be affected by the obstacle and the multi-path effect. These effects cause the strength of Wifi signal in a fluctuating state for a long time. If we use signal strength of single acquisition, it will produce a bigger error of positioning. In order to solve this problem, we use Gauss model to deal with signal strength of multi group in the sample period, reserve the signal strength of large probability and remove the signal strength of small probability.The experiments have proved that this method have improved the accuracy of signal acquisition and the accuracy of positioning systems.(2) The traditional KNN algorithm has low computational complexity and easy to implement, but its error is relatively bigger. The traditional Bayes algorithm has higher positioning accuracy, but its computational complexity is also higher. In order to meet the accuracy requirements and reduce the computational complexity effectively, this paper uses the methods of segmentation and interpolation to fusion KNN algorithm and Bayes algorithm. In the process of segmentation, we use MAC filter and KNN algorithm for rough positioning to narrow the scope of the positioning region and reduce the complexity of the matching computation; in the process of interpolation, we use the methods of horizontal interpolation and vertical interpolation to improve the fingerprint density in the localization area and improve the positioningaccuracy of Bayes algorithm.Finally, we design and implement a fingerprint positioning system based on the above theory of improvement and Android+J2EE+MySQL development platform. We have carried out the localization experiments in the real scene. The experimental results show that: The positioning accuracy of the algorithm proposed in this paper is slightly lower than the Bayes algorithm and much higher than the KNN algorithm, but the computational complexity is far lower than the Bayes algorithm. The algorithm proposed in this paper not only can meet accuracy requirements, but also reduce the computational complexity effectively. It is important to promote the development of indoor positioning technology.