Study On Parameter Estimation Algorithms Of Polarization Sensitive Arrays Based On Electrically Large Antennas And Compressive Measurements

Polarization Sensitive Array(PSA),also known as Electro-Magnetic Vector Sensor(EMVS)array,has recently received wide attention from academia and industry because of its advantage of polarization diversity.In the past two decades,many research works on PSA multi-dimensional parameter estimation have been achieved in academia and also have received essential applications in industry.However,some things could still be improved in the existing research: Firstly,the existing literature is mainly based on the PSAs composed of short dipoles and small circular loops.These arrays are electrically small antennas with low radiation efficiency,which may cause signal parameter estimation performance degradation for traditional algorithms.In recent years,long dipoles and large loops have been introduced into the field of PSA by their better radiation characteristics.However,the corresponding signal parameter estimation algorithms are relatively little studied.Secondly,to meet the demand of array systems for higher degrees of freedom and higher target information dimension,EMVS elements gradually replace scalar array elements,and the number of array elements is gradually increased,which will cause the number of front-end chains to surge and thus increase the hardware cost significantly.At the same time,the higher number of channels and higher information dimension will also increase the computational complexity of the algorithm.In response to these problems,a series of studies are conducted in this paper,and the main research contents and innovations are as follows:1.The measured data processing and analysis are carried out for an 8-dipole polarizationsensitive toroidal array.Firstly,the time domain echo signals received by the dipoles with different polarization directions at different operating frequency bands are analyzed,verifying the PSA’s polarization diversity characteristics.Secondly,the Reduced-Dimension MUltiple SIgnal Classification(RD-MUSIC)algorithms are applied to the array to obtain the 2D-DOA pseudo-spectral search results,which verified the effectiveness of the PSA parameter estimation algorithm.2.Research on signal parameter estimation algorithms of the PSA composed of long dipole and large loops.On the one hand,to address the problem of poor radiation efficiency of short dipoles and small loops,a new algorithm based on Spatially Collocated and Electrically Large EMVS(SCEL-EMVS)is proposed.It utilizes the real-imaginary part relationship to eliminate the electrical size coefficients and formulate a group of equations for 2D-DOA and2D-RPA.Multi-dimensional parameters can be obtained by solving the group of equations.Then,computer simulation results verify the effectiveness of the proposed algorithm;comparative experimental results illustrate that the proposed parameter estimation algorithm can estimate 2D-DOA and 2D-RPA simultaneously compared with existing algorithms,avoiding the impact of a priori errors on the parameter estimation performance.On the other hand,considering the low number of estimable targets of a single SCEL-EMVS,low effective aperture of the array,and mutual coupling of the antennas,a sparse planar array structure based on Spatially Separated and Electrically Large EMVS(SSEL-EMVS)is designed,for which a new estimation algorithm based on two-steps defuzzification is proposed.First,apply the Estimation Signal Parameter Rotational invariance Technology(ESPRIT)to estimate the high-resolution but ambiguous 2D-DOA;Then,for two scenarios,i.e.,with or without known electrical size coefficients,use the vector cross-product algorithm and the single SCEL-EMVS-based parameter estimation algorithm to calculate the coarse(but without periodic ambiguity)estimation results;After that,by comparing the two estimation results,obtain the high-accuracy 2D-DOA estimation without any ambiguity.The computer simulation results illustrate that the proposed algorithm can provide more decent target numbers and higher accuracy 2D-DOA for more application scenarios if the electrical size is unknown/known.3.Research on signal parameter estimation algorithms of the PSA that is based on compressive measurements.To address the problems of a large number of EMVS array elements,complex hardware system,and high computational complexity of parameter estimation,a uniform planar array structure of EMVS is designed based on compressive measurement,which effectively reduces the number of front-end chains by inserting a sparse compressive network between EMVS and front-end chains.Based on this,a compressed RD-MUSIC(C-RD-MUSIC)algorithm is proposed.First,This algorithm obtains the high-accuracy estimation results of 2D-DOA by the two-dimensional global search,then solves 2D-RPA.To address the problem that the parameter estimation performance degrades due to randomly compressed networks,a sparse compression matrix optimization algorithm is proposed,which solves the optimal sparse compression matrix by transforming the sparse matrix optimization problem into a Rayleigh-Ritz problem.Computer simulation results show that the proposed compression array framework and signal parameter estimation algorithm can provide highaccuracy parameter estimation while reducing hardware cost,and the proposed optimization algorithm can improve the parameter estimation performance.4.In order to further reduce the computational complexity of the algorithm and expand the application scenarios,a new compressed and arbitrarily spatially distributed EMVS array is constructed,which effectively reduces the number of front-end chains by inserting a sparse phase-shifter compression network between the EMVS and the front-end chains.Based on this,a compressed ESPRIT-like algorithm(C-ESPRIT-like)is proposed.This algorithm can use limited factors with rotational invariance to estimate coarse 2D-DOA;then,use this coarse estimate result as the initial value of the C-RD-MUSIC algorithm to obtain the high-accuracy 2D-DOA by 2-dimensional local search.For the optimization problem of sparse compressed matrices,a mapping method from sparse matrices to non-sparse vectors is proposed,which firstly transforms the original optimization problem into a quadratic constrained quadratic programming problem and then obtains the optimal sparse compression matrix by using a semi-positive definite relaxation technique and a Gaussian randomization method.Computer simulations demonstrate that the proposed array,parameter estimation,and optimization algorithm can reduce the number of front-end chains and obtain high-resolution multi-dimensional parameter estimates more efficiently.The sparse compression matrix proposed in this paper can be implemented by a sparse analog phase shifter network,which applies to arbitrarily distributed EMVS arrays and is easier to implement in engineering.