Research On Node Operations In Wireless Localization Networks

Wireless localization network is always a focused topic of researchers at home and abroad,and has a variety of applications in 5G,Internet of things(Io T),wireless sensor network(WSN),and space-air-ground integrated network(SAGIN).The most important purposes of wireless localization network is improving the localization accuracy.However,in many specific cases,the constrained usage of multi-domain resource is the main reason to limit the localization accuracy,such as the number of nodes,the transmit power,the deployed area,etc.Nowadays,in time difference of arrival(TDOA)and angle of arrival(AOA)-based localization scenario,there are plenty of research works focusing on the localization parameter estimation and localization algorithm.Nevertheless,the selection,deployment,and power and bandwidth allocation are not paid more attention in recent literature.To improve the localization accuracy and reduce the energy consumption of localization network,it is urgent to investigate the theoretical research of node operation and optimization strategies.This thesis studies the node operations in both active localization and passive localization and focuses on three main research directions,i.e.,node prioritization,node selection,and node deployment.We propose the optimization framework of node selection in TDOA-based passive localization and tracking scenarios,and propose the optimization framework of joint node prioritization and deployment in TOA-based active localization scenario.Then,based on semi-definite programming(SDP),discrete monotonic optimization(DMO),genetic algorithm(GA),and other optimization techniques,we propose a series of node operation algorithms in wireless localization network and provide plenty of simulation results to validate the proposed algorithms.The main contributions of this thesis are summarized as follows.1.The node selection with a single Boolean vector in TDOA and AOA-based passive localization scenario is investigated,which is then regarded as the fundamental framework of the node selection in wireless localization network.First,the TDOA and AOA measurement equations are transformed as pseudo linear equations.Then,the weighted least squares(WLS)-based closed-form localization algorithm is proposed and the corresponding covariance matrix of the localization error is regarded as the metric to measure the localization accuracy for the selected nodes.Second,by considering the features of TDOA and AOA that each node can obtain three measurement equations and giving the reference node,we introduce a Boolean vector and expand it as a high dimension vector to select three measurements determined by one node simultaneously.Third,with the constraint of the number of available nodes,we formulate the nonconvex optimization problem,where the objective is the covariance matrix of the localization error determined by the selected sensors and the decision variable is the expanded Boolean vector.To solve this problem,we introduce several auxiliary matrices and utilize convex relaxation techniques to transform the original problem as a SDP,which can be solved effectively by the interior point method.Simulation results show that the proposed algorithm performs better than K-nearest method and is close to the exhaustive search method.2.In the presence of non line of sight(NLOS)and correlated measurement noises,the node selection with two Boolean vectors in TDOA-based passive localization is investigated.First,we derive the Cram?er Rao lower bound(CRLB)in three scenarios,i.e.,LOS,PSU-NLOS,and PSK-NLOS.Second,two Boolean vectors are introduced to determine separately which node is selected as the reference and which nodes are selected as the ordinary nodes.Meanwhile,the CRLB determined by selected nodes is regarded as the objective to formulate the nonconvex optimization problem,where the constraint is the number of selected nodes.Third,we propose the node selection algorithms based on convex optimization and greedy algorithms.On one hand,the original nonconvex optimization problem is relaxed as SDP which can be solved effectively by convex optimization techniques.On the other hand,two heuristic algorithms,i.e.,best option filling(BOF)and iterative swapping greedy(ISG)are proposed,which have a lower computational complexity compared with SDP-based algorithm.The main idea of these two algorithms is that the localization accuracy may be improved when new nodes are added or swapped,so that the optimal node subset can be attained iteratively.Numerous simulation results are provided to verify the proposed node selection algorithms.3.The node selection with two Boolean vectors in TDOA-based tracking scenario is investigated.Different from the localization scenario,position estimations in a period are given in tracking scenario.Our goal is to select the nodes with optimal geometrical relations to tackle with tracking while the number of selected nodes is not given.First,since measurements in the past period can be regarded as the prior information to improve the position estimation at current time step,based on Bayesian estimation theory,the conditional posterior CRLB(CPCRLB)is regarded as the metric to mea-sure the position estimation at each time step.Second,the multiobjective optimization problem is formulated to balance the tradeoff between the tracking accuracy and the number of selected node at each time step.Third,to avoid the performance loss caused by the relaxation operations of the SDP-based algorithm,we introduce the definition of discrete monotonic optimization(DMO)and propose polyblock outer approximation(POA)algorithm to solve the optimization problem.Simulation results validate that the formulated problem can balance the conflicting objectives and verify that the POA-based algorithms perform well than the SDP-based algorithms.4.The joint node prioritization and deployment problem is investigated in TOA-based active localization scenario.By finding the optimal position and the optimal power of unmanned aerial vehicles(UAVs),space-air integrated localization network(SAILN)focuses on improving the localization accuracy for the entire area of interest(Ao I).Thus,the joint position and power optimization for UAVs is the combination of node prioritization and node deployment techniques.First,we introduce the definition of squared position error bound(SPEB)to measure the localization accuracy,and then define the average localization accuracy increment(ALAI)to measure the average localization accuracy increment when extra UAVs are added into the original satellite network.Then,with the drone-based station to ground(D2G)model,the ALAI is regarded as the objective to formulate the joint position and power optimization(JPPO)problems in both static and dynamic SAILNs where the constraints are UAVs' deployment area,non-flight-zone(NFZ),and total available power.Third,based on genetic algorithm and convex optimization techniques,we propose the pure genetic algorithm(PGA)and the power reallocation two step algorithm(PRTSA)to solve the JPPO problems.Compared with three practical methods,the solution solved by the proposed algorithms can provide a better localization accuracy for the entire Ao I.