Signal Strength Feature Information Fusion For Indoor Wi-Fi Localization Based On D-S Evidence Theory

With the widely arrangement of indoor various Wi-Fi terminals,indoor Wi-Fi location method based on Received Signal Strength(RSS)has become one of the attractive topics in the area of mobile communication.However,the available schemes rarely take into account the influence of signal distribution diversity on signal propagation distance estimation due to the blocked information transmission and multipath effect,and the influence of RSS information uncertainty on system performance,which both lead to large error and terrible robustness of the localization system.Aiming at addressing those deficiencies,a signal strength feature fusion method for indoor Wi-Fi localization based on the Dempster-Shafer(D-S)evidence theory is designed,and the main research contents are as follows.Firstly,the heuristic distribution model is used to establish the mathematical relation between RSS and signal propagation distance from different Wi-Fi Access Points(APs),and the upper and lower bounds of signal transmission distance offset based on the signal distribution uncertainty are built,based on which the distance distribution uncertainty is programmed into the measurable range of probability,so as to reduce the distance estimation deviation.Secondly,the kernel density estimation method is exerted to work out the signal propagation distance distribution evaluation under different APs and signal strength.Specifically,according to the difference of signal propagation distance distribution under different RSS values,the characteristic function of signal propagation distance distribution based on RSS is constructed by kernel density estimation,and the normalized estimation of signal propagation distance distribution is taken as the basic probability assignment of the D-S evidence theory.Finally,the indoor positioning system is modeled by the D-S evidence theory.Specifically,first of all,each AP is regarded as an independent information source,and the identification framework of indoor positioning model is established.Second,the signal features of multiple AP information sources are fused according to the evidence composition rules of the D-S evidence theory,and the matching Reference Points(RPs)sets for targets are constructed by trust function.Finally,the ideal matching RP of the target is constructed through the decision rules of the trust function to realize the precise estimation of the target position.The extensive simulation test results demonstrate that the proposed scheme has higher localization accuracy and stronger system robustness.