Direct Emitter Localization Technique Based On Spatial Spectrum Estimation

From short wave to millimeter wave,from over-the-horizon detection to wireless sensor network,from military to civil purposes,the high accuracy localization of emitters is imminently and widely required in various of applications for different frequency bands,different propagation environment,and different system structures.Traditional indirect emitter localization methods ignore the internal relations between the measurements,and their data association procedures bring uncertain effects to the estimation results,thus their performance are limited in the practical applications.Benefiting from the properties of high accuracy and super resolution,the spatial spectrum estimation techniques have been widely used for measuring the incident angles of the received signals.Besides,the spatial spectrum estimation techniques are also well developed according to different requirements and situations.Following the idea of direct localization,focusing on the problems of system and algorithm design,this thesis introduces several excellent approaches of spatial spectrum estimation with expansion and modification,aiming to achieve the enhancement of emitter localization performance.The main contents and contributions of this thesis are given below:Firstly,to improve the estimation accuracy of emitter localization systems,this thesis proposes a design method for directional antenna arrays and a calibration method for non-ideal arrays.In the design method for directional antenna arrays,Cramér-Rao bound(CRB)based optimization models are established by utilizing the least squares fitting technique,aiming to achieve the optimal radiation characteristics which can improve the performance of angle estimation in a predetermined objective spatial sector including all the potential directions of incidence.Besides,a modified simulated annealing(SA)algorithm with the iteration of parameters is proposed,aiming to solve the optimization problems when the classic SA is invalid.Compared with the corresponding conventional ULA,an optimized array can obtain higher accuracy of DOA estimation in the objective spatial sector with little fluctuation.Additionally,the optimized design of radiation characteristics can also suppress the ambiguities,and remains effective for the arrays with different aperture.In the calibration method for non-ideal arrays,this thesis modify theclassic manifold separation technique(MST),aiming to reduce its dependence on high signal-to-noise ratio(SNR)measuring environment.According to the analysis of the array response,it is demonstrated that to maintain a correct phase relationship between the received data at different calibration angles is indispensable for the application of MST.Thus,this thesis slightly change the structure of the traditional calibration system,so that a phase reference for the measurements can be obtained.Besides,unlike the classic MST,where only a single snapshot measurement is utilized for calibration,multi-snapshot information is exploited in the novel method by using the subspace decomposition technique.Compared with classic MST,the proposed method can achieve better calibration performance with the same environment,which will lead to higher estimation accuracy.Secondly,to solve emitter localization problems for the classic linear model,this thesis proposes two angle-of-arrival(AOA)based direct localization algorithms for global and local narrowband conditions,and a direct localization algorithm jointly using time-difference-of-arrival(TDOA)and AOA information.On the global narrowband condition with AOA information,we propose a novel localization model which has the form of multidimensional harmonic retrieval based on spatio-temporal processing.By utilizing multidimensional spectrum estimation techniques,the underdetermined localization problem can be handled thanks to the increase of degree of freedom.To further improve the identifiability,a nested array based localization model is also given based on the multidimensional processing framework.The proposed NM-Capon algorithm can achieve higher localization accuracy without data association and the prior knowledge of model order,and is robust to the correlated noise.On the local narrowband condition with AOA information,this thesis derive the maximum likelihood(ML)estimator,and solve the optimization by applying the expectation-maximization(EM)and polynomial rooting techniques.The proposed method does not require the data association between stations,can achieve better localization performance than the methods on the same condition,and is more robust against signal bandwidth when compared with the methods designed for the global narrowband condition.For the model with TDOA-AOA information,this thesis proposes a direct localization algorithm based on spatio-temporal processing.The algorithm formulates the time-delay cross-correlation matrices with the received data of each two stations,and localize the emitters according to the properties of the matrices andthe subspace based techniques.The proposed method does not require data association and can work in the underdetermined scenario.Compared with the AOA based methods,the proposed method has better performance in the scenario of low signal-to-noise ratio(SNR)or large bandwidth.Compared with existing TDOA-AOA based methods,the proposed method has better performance in the scenario of high SNR or small bandwidth.Finally,a novel perspective for the problem of direct localization of emitters is presented based on sparse Bayesian learning(SBL).The localization problem is addressed in three scenarios,where this thesis derive the AOA based models for global and local narrowband conditions and the model jointly taking use of TDOA and AOA information.Unlike the existing ?1-penalization based direct localization methods,this thesis expand the localization models to the sparsity based framework with the scheme of SBL,which makes the partly unknown dictionary based problems solvable.On the basis of the boundoptimization SBL algorithm,modifications are provided to make its solving procedures adapt to the localization models.For one thing,the learning rule of channel attenuation factors is added in the solving procedures.For another,this thesis expand the original single measurement vector(SMV)based model to the multiple measurement vector(MMV)scenario.Furthermore,this thesis modify the original procedures by updating the parameters with an alternating minimization strategy,which guarantees the convergence of the algorithms.The proposed SBL based methods does not require data association,the prior knowledge of model order and the choice of hyper-parameters,and can achieve higher of similar localization accuracy compared with existing methods.