Research On LTE Fingerprint-based Positioning Algorithm

In recent years,with the rapid development of Information Techonology and Communication Technology and the rapid popularization of various intelligent devices,the 5th Generation communication(5G)system revolutionary put forward the concept of "Internet of Everything".Location based services are attracting more and more attention,and will be widely used in the popular fields of "Internet of Things","Internet of Vehicles" and so on.Thus,high-accuracy positioning,especially high-accuracy outdoor positioning,has become the focus of more and more researchers.With the rapid development of 4th Generation communication(4G)system,urban areas have been widely covered with long term evolution(LTE)signal,which makes it possible to use LTE positioning navigation systems on land as an alternative to Global Navigation Satellite System(GNSS).3GPP Release 10 proposes a Minimization of Drive tests(MDT)technology,which are reported by the LTE user equipments(UEs)including reference signal receiving power(RSRP)and exact location,which makes the construction of radio map easier.However,the mobile diversity would lead to an inherent RSRP difference,what's more,different UE motion states will bring different RSRP fingerprint features.In this paper,based on the above problems existing in the LTE fingerprint-based positioning system,a RSRP difference elimination algorithm and and a UE motion state classification algorithm are proposed,which can effectively,improve online positioning accuracy.Firstly,aiming at the problem of involving inherent RSRP difference because of the mobile diversity in the process of constructing Radio Map by MDT reports,this paper presents a UE RSRP elimination algorithm based on Multidimentional Scaling(MDS).The RSRP difference for each UE is calculated through some position-known RSRP fingerprints by MDS algorithm.Then RSRP is eliminated both in the offline phase and the online phase.The proposed method can effectively eliminate for the inherent RSRP difference,so as to improve the positioning accuracy in the online phase.Secondly,different UE motion states lead to different fingerprint features.Still state has a short fingerprint and fingerprint misalignment is serious for moving state.This paper prosents a UE motion state classification algorithm.Radio Map is categoried into a moving one and a still one.In online,we also make classification on RSRP motion state to match with the corresponding radio map.The proposed method can exhibit the advantages of fingerprints in separate motion states and improve positioning accuracy.