Research On Distributed Adaptive Target Localization Algorithms Based On Wireless Sensor Network

Multiple targets localization has been attracting much attention in the field of signal processing and widely applied in radar,sonar,navigation,radio-astronomy and seismol-ogy.In recent decades,direct position determination(DPD)has become a major topic due to its advanced performance in specific scenarios.As the number of studies on distributed wireless sensor network increases,it is obvious that the algorithms could benefit from the expansibility and flexibility of the distributed algorithm.The problem of how to make different equipments in a heterogeneous network work together has drawn much attention due to the rapid replacement of hardware nowadays.It is worth exploring to solve the target localization problem in a distributed manner using heterogeneous network.This thesis focuses on directly locating multiple targets using distributed heterogeneous array networks,the main contributions are summarized as:A distributed adaptive node-specific multiple targets direct localization algorithm us-ing a static heterogeneous array network is studied in this thesis.Meanwhile,a transformed-dimension scheme tailored for the heterogeneous array network to make the different ar-rays locate the targets together is studied in this thesis.Based on the minimum variance distortionless response(MVDR)criterion and the gradient descent method,the compu-tational complexity caused by the massive grid search in conventional DPD algorithm is avoided.A scheme to adaptively estimate the inversion of the received signal correlation matrix based on the matrix inversion lemma is also discussed in this thesis.The scheme makes the studied algorithm able to continuously update the position estimates of slow-moving targets.A high dimensional signal correlation matrix direct localization(HDDL)algorithm is studied in this thesis for the signal model where the transmitted signal is assumed to be known.The HDDL could achieve better performance in terms of localization accu-racy at the cost of a high dimensional correlation matrix,which might lower the conver-gence speed of the HDDL.Based on the HDDL,a reduced dimensional direct localization(RDDL)algorithm is studied.And the joint iterative scheme is implemented to simulta-neously calculate the position estimates along with the reduced dimensional matrix.Considering the case where a moving array network is used to locate multiple static targets,a centralized direct localization for moving array network based on grid search(CDLMG)is studied with the Doppler frequency taken in to consideration.Based on gradient descent method,the CDLMG is also extended to a distributed scheme in this thesis and the distributed version is named as DDLMG.Herein the transformed-dimension scheme is further adjusted for the moving hetero-geneous array network.The relationship between the steering vectors in two adjacent time intervals.As is illustrate in the simulations,both the CDLMG and DDLMG are able to complete the mission of locating multiple targets.