The Development Of Indoor Positioning System Based On Deep Learning From Wi-Fi Fingerprint Data

With the unceasing development of human society,cities will bear more and more people in the future.At present,our country is in the period of accelerated urbanization.To solve the problem of urban development and achieve sustainable development of cities,building smart city has become an irreversible trend of development of world cities.In the construction of a smart city,intelligent building is an important part of smart city,and has great significance for the construction of smart city.In the process of highly urbanized development,people's indoor activity space is becoming more and more complex and huge,and the demand for positioning and guidance in indoor parking,shopping malls and airports is becoming more and more intense.The demand for indoor positioning has been unprecedented.This is also the demand for smart buildings and smart city construction.Global Position System has high performance and accuracy,but it can't penetrate the buildings and carry out indoor positioning.Indoor positioning technology is developing rapidly as the development of personal devices such as WLAN and smart phone.Wi-Fi wireless networks have covered the streets of big cities.Combined with the technical characteristics of Wi-Fi,many scholars have studied the indoor positioning technology of Wi-Fi,and put forward many the position of the fingerprint algorithms.But the positioning performance of traditional positioning algorithms still needs to be improved.In this paper,we use data driven to develop an indoor positioning system based on Wi-Fi fingerprint big data and deep learning,aiming at the problem of interference and positioning accuracy.The specific research work is as follows:(1)The deep learning method,Autoencoder,is used to automatically extract the depth features of Wi-Fi fingerprint big data,and create a one-to-many relationship fingerprint database using depth features.(2)The technology of Wi-Fi indoor positioning is studied.The reason why Wi-Fi technology is used for indoor positioning is expounded,the traditional indoor location algorithm is improved,and a count-k nearest neighbor indoor positioning algorithm based on the depth feature fingerprint database is proposed in this paper.The design idea and implementation of the algorithm are described,and the algorithm is simulated.(3)The indoor positioning system is developed.System client is based on Android.Using Java to complete the development of server Web Server in the integrated development environment of MyEclipse.The algorithm is implemented in Python under the Anaconda.The system has been tested in the parking lot of Chongqing ASE center square.The feasibility and accuracy of the system in the complex indoor environment are proved by the result.It also shows that the system has practical significance.