Research On Array Antenna-based Adaptive Beamforming Algorithm

Recently,array processing technique is a hot topic for domestic and overseas researchers.By using a sensor array of a specified configuration to form a multiple antenna system,array processing can have a good performance for parameter estimation when received signal vectors contain interference and noise.This paper mainly focuses on beamforming problem which is also called spatial filtering technique.Beamforming can take advantage of the sensor array to enhance desired signal component and suppress interference and noise.As beamforming has a strong interference rejection and gain processing capability,it is widely used in radar,sonar,wireless communication,GPS navigation,astronomy and medical image.Weight vector design is the core content of adaptive beamforming.There is different prior information in different applications.Considering those information,we can choose distinct design criteria obtaining optimal weight vector.Taking design criteria as links,several adaptive beamforming algorithms and their improved versions are proposed in this paper based on different design criteria,in order to increase convergence speed and tracking performance,enhance robustness of beamformer and reduce installation cost.The main work and innovations of this dissertation are summarized as followed:Firstly,in order to improve the convergence speed of fixed forgetting factor Recursive Least Squares(RLS)based on the minimum mean square error(MMSE)criterion,we proposed a variable forgetting factor RLS and applied it to widely linear adaptive beamforming.By using the relationship between the noise-free a posterior and a priori error signals,the optimal forgetting factor expression can be derived,which guarantees the optimality of the forgetting factor at each snapshot.Aiming at high computational cost of the proposed widely linear variable forgetting factor RLS algorithm,a low computational version is proposed by exploiting the structure of the augmented covariance matrix.Compared with the former,the latter not only has a better convergence speed,but also owns a lower computational complexity.In simulation part,the effectiveness is verified in terms of computational complexity and output Signal to Interference plus Noise Ratio(SINR).Secondly,under the constant modulus(CM)criterion,three steering vector of desired signal constrained UKF algorithms are proposed and applied to adaptive beamforming.The first proposed algorithm operates in Direct Form(DF)structure and uses the iterative projection method to design DF-based constrained constant modulus UKF(DF-CCM-UKF)algorithm.The second proposed algorithm makes use of Generalized Sidelobe Canceler(GSC)structure to transfer the constrained optimization problem to unconstrained optimization one.The constant modulus cost function is converted to a state space model where we can use UKF to obtain the weight vector of adaptive beamformer,developing proposed GSC-CCMUKF algorithm.However,compared to least mean squares(LMS)and RLS,UKF has a higher computational complexity.In order to solve this issue,a krylov subspace-based reduced-Rank(RR)method is introduced to GSC-CCM-UKF algorithm,i.e.RR-GSC-CCMUKF.The improved algorithm has a faster convergence speed and lower computational cost than GSC-CCM-UKF.In simulation part,proposed algorithms are compared with existing RLS-based adaptive beamforming methods to demonstrate their effectiveness.Thirdly,in practice,there are many inaccuracies in sensors array,such as desired direction mismatch,array position error,wavefront distortion.Since these factors can lead to serious performance degradation of adaptive beamformer,we focus on the robustness of beamforming.There are numerous research results for processing Gaussian signals combined with minimum variance(MV)criterion.However,non-Gaussian signals-based adaptive beamforming methods are not studied adequately.Therefore,a novel constant modulus robust algorithm with probability constraint is proposed.By utilising statistical information of a priori error and constant modulus feature of a communication signal,the proposed algorithm can achieve a better robustness than existing methods which could be verified through simulation results.Finally,among linear sensor array configurations,sparse array can have a better parameter estimation performance than a uniform linear array.Compressive sensing and subspace technique are respectively used to design two novel coprime array-based adaptive beamforming algorithms.For proposed sparsity-based adaptive beamforming method,direction and power estimation problem is converted to a compressive sensing problem which could be solved by using Lasso.As Lasso has a feeble power estimation capability,we proposed an improved version by using the least square method.In order to suppress interference and noise effectively,those estimated DOA and power information is used to reconstruct the interference-plus-noise covariance matrix designing weight vector of beamformer.For proposed subspace-based adaptive beamforming method,the covariance matrix of the virtual received signal and a matrix through integrating spatial spectrum are eigen-decomposited to form an overlap subspace which could be used to estimate the signal steering vector of the virtual array.Then,the signal and noise power can be estimated by utilizing modified Capon power estimation methods.We make use of those estimated information to design adaptive beamformer on physical coprime array.In simulation part,the proposed algorithms are compared with existing similar type of methods to demonstrate their effectiveness.