Research On 3D Passive Source Localization Methods For Wireless Sensor Networks

Target intrusion detection is one of most important application Wireless Sensor Networks(WSN).Location information of target is very important for monitoring events.Therefore,positioning technology needs to be used to determine the exact position of unknown intruding objects.At present,the method of target location estimation in WSN is mainly based on the measurement angle of arrival(AOA),time difference of arrival(TDOA)and gain ratio of arrival(GROA).However,these direction and distance related measurements are highly nonlinear,and it is related to the geometric location of WSN nodes.How to accurately obtain the location information of passive intruding targets is a challenging work.In this thesis,we summarize the current research of passive target localization of WSN,and analysis the TDOA location with multiplicative noise and AOA-GROA positioning with uncertain sensor location respectively.The main contents of this thesis are as follows:1.A semi-definite relaxation optimized TDOA localization algorithm is proposed for the localization with multiplicative noise.This algorithm can avoid the problem that the traditional maximum likelihood(ML)estimation algorithm is easy to diverge.Firstly,the ML equation about multiplicative noise TDOA is obtained and the ML problem is reconstructed to obtain the optimized expression.Secondly,the semi-definite relaxation is used to make optimized expression transformed into convex optimization problem,and solve it by CVX Convex optimization.In the simulation,we have compared respectively for two different measurement models,additive noise and multiplicative noise.The ML method and the least squares method are chosen as contrast algorithms to analyze the localization performance of the above algorithm under different signal-to-noise ratios.The simulation results verify the proposed method.2.A AOA-GROA constrained least square hybrid localization algorithm is proposed for the localization with uncertainty of sensor location.Compared with the TDOA localization algorithm,this method does not require time synchronization.Firstly,the pseudo-linear equations of AOA and GROA are established to obtain the weighted least-squares estimation of the target position.Then,the constrained least-squares estimation solution of the target position is obtained through the optimal solution of the cost function with the WLS estimation.The theory performance of purposed algorithm is also deduced.Through simulation experiments,the influence of different sensor arrangement and different standard deviations of AOA error,GROA error and position error on positioning results was analyzed and compared.The simulation results show that the joint position estimation of AOA-GROA is related to the relative geometric position of the sensor.Compared with the direction localization algorithm,the performance of proposed algorithm is better.