WIFI Positioning Algorithm Based On Deep Learning

WIFI positioning technology is an important branch of wireless location, has been widely used on many important occasions, such as hospitals, museums, shopping malls and so on. But currently WIFI positioning accuracy problem remains limits its application. This paper for the WIFI positioning issues, first introduce the background knowledge and relevant technologies, then introduction the new methods of new areas, to improve WIFI positioning accuracy.This paper first introduces the neural network which has highly non-linear modeling capabilities, starting introduce the most basic BP neural network, and then describes the integrated approach, which can further improve positioning accuracy based on the existing neural network. Then we described the auxiliary optimization methods for the neural network.Finally, this paper describes the new technology of RBM, DBN and SGRBM and use for WIFI positioning, these technologies can improve the positioning accuracy on the basis of original BP neural network. At last the experimental vertification stage, we collected data in the real scene used for compare different methods of positioning accuracy, the result is that the positioning accuracy of BP neural network to improve the capacity of the existing related technologies more than10%, the proposed method is based on DBN BP network, the further improvement of positioning accuracy.