Research On Indoor Wireless Location Based On FWA-SVM

With the rapid development of mobile Internet and intelligent terminals,we have entered the era of mobile information.We need more LBS by wireless network than before.However,the most of current LBS are only available outdoor.In recent years,many scholars have researched on indoor positioning methods,which still have some problems,such as unstableity,low efficiency and positioning accuracy.Aiming at solving these problems,we propose an improved SVM by Fireworks Algorithm as the indoor postioning method,which is better than two typical location method,K-nearest Neighbor Algorithm and Neural Network.Moreover,the experimental results show that the proposed method has good efficiency and accuracy in indoor location.The main research are as follows:(1)Apply the Kriging interpolation method to expand the fingerprint database.In the fingerprinting location method,we need to set up a fingerprint database offline,which will take much more work in a large area.So we apply the Kriging interpolation method to solve this problem.And the final experiment shows that it can reduce 40% work.(2)Use SVM classification and regression method to locate.We find that only using SVM regression method is not suitable for the location in large area,which will take long time for location.Therefore,we classify the area with different parts,then apply SVM classification to get which part does the user in,and obtain the final location by SVM regression method.This can not only effectively reduce the positioning error,but also shorten the positioning time.(3)Propose FWA-SVM location method.In SVM classification and regression method,the selection of penalty parameter and kernel parameter play an important role.In this paper,we choose a new swarm intelligent algorithm-the fireworks algorithm to optimize these parameters.Compared with another typical swarm intelligent algorithm-Particle Swarm Optimization(PSO)by experiments,this proposed algorithm has better searching time and effect.