Research Of Array-based Direct Position Determination Technology

Wireless localization technology plays an important role in both military and civil fields.Lots of localization algorithms have been proposed in recent years.The most common methods for position determination of radio signal emitters are based on measuring a specified parameter such as angle of arrival,time of arrival or frequency difference of arrival of the signal in the first step.The measured parameters are then used to estimate the emitters' location in the second step.These localization methods can be called two-step localization methods.The processing of two-step localization methods is relatively simple,while it is inevitable to cause location information loss because of separation of parameter estimation and position calculation.Thus these traditional twostep localization methods can not achieve the optimal estimation performance.Direct position determination(DPD)technology can locate emitters directly from the raw observation data without parameter estimation processing.Therefore,DPD methods can gain higher localization accuracy than traditional two-step localization methods.And it has become a research hotspot in wireless localization domain nowadays.Taking localization accuracy and computational complexity as targets,this paper developed the research of array-based DPD algorithm with the support of the key military project.The main work and innovation of the paper are summarized as follows:1.Traditional array-based DPD methods can not exploit Doppler frequency shift information to improve localization accuracy.To overcome the shortcoming mentioned above,an array-based DPD method fusing Doppler frequency shift information is proposed.Firstly,an arraybased DPD model fusing Doppler frequency shift information is constructed.Then,a maximum likelihood(ML)estimator is designed and emitter's position estimation is transformed to calculating the maximum eigenvalue of the matrix containing location information.The property of eigenvalue keeping unchanged after matrix transpose is used to simplify the calculation.Finally,cramer-rao lower bound(CRLB)of this new DPD model is derived.Simulations show that the proposed method has higher localization accuracy than the existing DPD methods which based on array antenna model or only utilizing Doppler frequency shift information.2.Traditional DPD methods are based on the point source model and have localization accuracy decrease when locating distributed sources.Two DPD methods of distributed sources are proposed to overcome the above shortcoming.Firstly,a DPD model which can describe spatial distribution characteristics of emitters accurately is constructed.Next,based on ML criterion,a DPD method with optimal localization accuracy is proposed to locate a single distributed source.Based on the idea of eigenspace decomposition,a DPD method with relatively lower computational complexity is proposed to locate multiple distributed sources.Finally,the CRLB of the DPD model of distributed sources is derived.Simulations show that the proposed methods have higher localization accuracy than the traditional DPD methods when locating distributed sources.3.Most of existing DPD methods have huge computational complexity because of locating emitters via traversal grid search.A fast DPD method based on sharpening genetic algorithm(SFGADPD)is proposed to reduce calculation.Firstly,a fitness function based on the idea of sharpening processing is designed to enhance selection ability of excellent individuals.Then,the convergence strategy is optimized to decrease probability of falling into local optimum.Simulations show that,with the guarantee of localization accuracy,the computational complexity of the proposed method can be reduced 3 orders of magnitude at 95% probability.4.To accelerate calculation of SF-GADPD method,a parallel architecture of GPU is designed according to parallel features of fitness function and genetic operation of SF-GADPD method.Experiments based on GPU platform show that the proposed method has high computational efficiency and engineering application value.