Research On Cooperative Localization Algorithm For 5G Network

With the development of 5G mobile communication technology and the increasing number of networked devices,a large number of applications(such as emergency rescue,automobile navigation,intelligent transportation and industrial automation,etc.)need localization information with different positioning accuracy.Multiple Input Multiple Output(MIMO)systems,millimeter wave(mmWAVE)communications,Ultra-Dense Network(UDN),Device-to-Device(D2D)communication and other technologies brought by 5G network can be used improve the positioning performance.In order to improve the positioning accuracy of nodes under 5G network,it is a choice to introduce the cooperative localization algorithm into 5G wireless localization.In this dissertation,the technology of Multidimensional Scaling(MDS)and the selection of anchor nodes in cooperative positioning algorithm are studied,and the main work and innovation points are listed as follows:(1)In the positioning process of classical MDS algorithm,the distance between nodes cannot all be directly measured,and the shortest path estimation is generally carried out through multi-hop distance,where the positioning accuracy is greatly affected by the accuracy of distance estimation.Therefore,an MDS positioning algorithm based on asymmetric generalized centralized matrix is proposed.In this dissertation,a class of asymmetric generalized centralized matrices is proposed,and the influence of centralized matrices on positioning errors is studied.A centralized matrix which is helpful to suppress the "error" distance is selected to improve positioning accuracy.The experimental results show that,compared with the classical MDS algorithm,the MDS algorithm based on the asymmetric generalized centralized matrix breaks the limitations of the original centralized matrix,and can reduce the impact of distance estimation error on the localization and improve the positioning accuracy of nodes.(2)A distributed MDS algorithm based on improved centralized matrix was proposed to solve the problems of node distance estimation error in existing distributed MDS localization algorithm and error accumulation in local map merging process.Firstly,the algorithm clusters nodes according to the number of hops between nodes in the network.Then the corresponding centralization matrix is designed to double centralize the matrix with the square of the global shortest path distance,and the local relative coordinates are obtained.Finally,the error accumulation in coordinate synchronization process is suppressed by using iterative method to patch-and-stitch the local coordinate into a global one.The simulation results show that the proposed algorithm can not only suppress the influence of multi-hop ranging error,but also optimize the local map merge error through iterative method and improve the overall positioning accuracy by designing appropriate centralized matrix.(3)Aiming at the problem of improving localization accuracy by anchor node selection in cooperative localization,a localization strategy based on optimal equality constraint is presented.Using the Cramér-Rao Lower Bound(CRLB)as the evaluation criterion of positioning accuracy,the optimal equation constraint for the lowest CRLB trace under the given constraint conditions is derived,and the feasibility of constructing the optimal constraints before knowing the node ground truth locations is analyzed.In the simulation,the performance of anchor-type constraints and the optimal constraints is compared,and the number of equality constraints under different environments is given.Based on the research on the performance of MDS algorithm in cooperative localization,this dissertation proposes the positioning algorithm with improved centralized matrix and anchor node selection strategy,which improves the positioning accuracy of the classical algorithm,and contributes to promoting the application of cooperative localization algorithm in 5G network.