Research On RF Fingerprinting In LTE Network

With the rise of various location-based services (LBS) and the requirements for E-911 emergency positioning, wireless mobile positioning technology develops rapidly. Grid-based radio frequency (RF) fingerprinting is a database correlation method (DCM), which only relies on the feature of electronic signals to provide the positioning service. It is a simply positioning method and can overcome the interference of buildings. Thus, the research focuses on RF fingerprinting in last decade. The main challenges of RF fingerprinting are matching complexity, the scale of correlation database and the positioning accuracy. In addition, grid-based RF fingerprinting divides the target areas into a large number of rectangular grids and another challenge is the type of grids. In this thesis, the fingerprint matching algorithm and the mode of grids under the LTE (long term evolution) homogeneous and heterogeneous networks are studied.Firstly, a simplified algorithm of matching and two similarity metrics are proposed. One is the difference of Reference Signal Received Power (RSRP) sequence, and the other one is the variance of this sequence. In the training phase, collect the information using MDT (minimization of drive testing measurements) and construct the correction database. In the positioning phase, match the Cell-IDs of training signature and testing signature and perform the RSRP matching when the differences of Cell-IDs and the number of Cell-IDs are smaller than threshold. Otherwise, give up the training signature. According to two similarity metrics, calculate the similarity distances. The training signature who has the smallest similarity distance is regard as the matched signature, and the position of testing signature is obtained. The simulation results show that the simplified matching algorithm decreases the matching complexity at least by 34%, and guarantees the positioning accuracy.Secondly, according to the conventional regular grid layout, a meshing method based on uneven grid layout is proposed. In the coverage area of one base station, the signal strengths in different positions are calculated and the relationship between signal strength and positon is observed. Based on the rule of hexagonal, we divide the coverage area of one base station into multistage hexagonal areas and set different levels. The area near the base station is set a high level. At the edge of the base station, lower signal strength and lower distinguishability will decrease the positioning accuracy, and we set smaller grid size. Otherwise, we set bigger grid size at the area near the base station. The same division is set to the target area and the uneven grid layout is obtained. The simulation results show that this method efficiently decreases the scale of correction database and the positioning accuracy increases at most by 16%.Finally, a dynamic uneven grid layout is designed under the uneven grid layout. We set the basic boundary adjustment piecewise ratio and the extension ratio for all areas. By changing the ratios, we can change the distances from the side of hexagonal to the center in different level areas. Then the multistage hexagonal areas are adjusted. Simulation results show that the areas with smaller grid size extend, the scale of database will become bigger and the positioning accuracy will increase obviously. Otherwise, the scale of database will become smaller and the positioning accuracy will decrease obviously. This controllable boundary adjustment method not only guarantees the non-overlap of multistage areas, but also increases the flexibility of uneven grid layout.