Reseach And Implementation On WLAN-based Indoor-localization Techniques

With the increasing need of Location Based Service,Indoor-Localization has become a more important topic.Indoor-Localization algorithm aims at resolving the“one last kilometers ” problem,thus to resolve the problem that traditional navigation systems can not solve. Shelter from the walls,the GPS signal can not come across the rooms,which affects its accuracy.The new Indoor-Localization algorithm came out due to the reasons mentioned above.As Wireless Area Local Network has become a crucial part of modern life,access point has been everywhere,and the receive signal strength is also easy to measure and use.Using WLAN to execute Indoor-Localization has become an important direction.By the influence of various aspects,including environment changing,user moving,etc.,however, there is still a lot to be done.Basic on the reason,in this paper the main research is focused on how to improve the existing indoor location based on WLAN technology, at the same time, it was designed and implemented a set of indoor positioning system based on WLAN.First, this paper introduces the basic principle of indoor fingerprint localization based on WLAN technology, and analyzes the present research on indoor main difficulties of fingerprint positioning algorithm, at the same time introduces the mainly several kinds of the existing indoor localization algorithm at home and abroad,including K approaching method, support vector regression method, the primary difference between these algorithms is to set up RSS and positioning coordinate mapping methods, then analyzes the advantages and disadvantages of learning algorithm in indoor positioning system, for the later research and provides a theoretical basis.Next, this paper continue to analyze the learning algorithm is an important algorithm, the basic principle of neural network algorithm and how to apply in the indoor location, and neural network algorithm is analyzed through theoretical derivation some key points in implementation. And put forward a kind of ascension in the indoor positioning precision of the neural network algorithm method, namely the secondary training method. Then using the adjacent relation of neural network algorithm was further improved. At the same time, the use of the accuracy of neural networkclassification results under different fingerprint density difference, and then puts forward a regional judgment algorithm, and use the map matching algorithm for locating results correction, the last of the initial neural network algorithm and several improved algorithms were tested, and the results are compared and analyzed.Finally elaborated this topic how to design and implement an indoor positioning system, and analyzes the experimental environment and possible errors in the training data, at the same time set threshold value method is used to remove the peak error in the data in the training data.