Research On Broadband Beamforming For Digital Array

With the recent developments in antenna array technology and digital technology,the digital antenna array,especially narrowband digital antenna array,has been used extensively in many applications.The broadband digital array radar(DAR)using antenna array technology has higher resolution,better target classification capability,stronger target identification capability,better anti-jamming performance,improved reliability and superior capability to detect and track centralized targets.Broadband digital beamforming(DBF)is one of key technologies in broadband DAR.Due to spatial dispersion and aperture fill time,the broadband DBF with conventional phase shifter has large beam pointing errors over the passband.Therefore,new techniques for broadband DBF need to be developed.This dissertation is focused on broadband DBF technologies,including non-adaptive and adaptive broadband DBFs.The algorithms and methods proposed in this dissertation can be used for real-time applications with reduced complexity but with improved broadband DBF performance.The main contributions of this dissertation are as follows:1.Since broadband DBF needs digital true time delay(TTD)to ensure correct beam pointing,this dissertation explores the use of variable fractional delay filter,(i.e.,Farrow filter),for broadband DBF.The algorithm method is derived and analysed in detail.The broadband DBFs using floating-point and fixed coefficients Farrow filters are implemented using MATLAB and we verify that the DBFs using Farrow filter yield accurate beam pointing.The hardware realizations of the Farrow filter and broadband DBF using Farrow filter are presented.The evaluation of DBF using the Farrow filter is presented based on existing real-time signal processing systems.The DBF using Farrow filter is verified on FPGA hardware systems and we show that it has the correct beam pointing and reliable beampattern performance.Those results confirm that the Farrow filter can be used for accurate and effective broadband DBF.2.Since the existing frequency invariant beamforming(FIB)cannot yield anti-interference nulling,the constraints of pattern nulling and the prototype equiripple finite impulse response(FIR)are respectively added to the original Fourier transform FIB to obtain patterns with low sidelobe and anti-interference nulling.In addition,to reduce the computational complexity of the FIB,the sparse optimization of the weights of the FIB is proposed based on minimum l₀-norm.The optimization is solved by the orthogonal matching pursuit(OMP).To reduce the number of tapped delay lines(TDLs),the sparse optimization of the TDLs is also presented,which effectively decreases the number of sensor elements and reduces the implementation complexity.The simulation results verify the correctness and effectiveness of the proposed methods.3.Since the broadband spatial-temporal beamforming algorithms have high computational complexity,the low-complexity algorithms are proposed for spatial-temporal beamforming in this dissertation.For a discontinuous wave broadband DAR,a generalized reduced-rank structure for spatial-temporal beamforming,which is based on linearly constrained minimum variance(LCMV),is presented to reduce the dimensionality of the received data vector,reducing the complexity of obtaining the adaptive weights.The singular value decomposition(SVD)operation for the frequency domain constraints is applied to the algorithm for release numerous degrees of freedom(DOFs)and to yield better anti-interference performance.Without the eigen-decomposition of the received data covariance matrix,a fast algorithm for building the reduced-rank matrix is presented in this dissertation to reduce the computational complexity of building the matrix.With undistorted response to the desired signal and satisfactory anti-interference capability,the proposed algorithm reduces the computational complexity of the adaptive weight approach.For a continuous wave broadband DAR,the generalized reduced-rank structure based on spatial-temporal generalized sidelobe canceller(GSC)is proposed.Based on the LCMV reduced-rank algorithm,a fast reduced-rank algorithm for spatial-temporal GSC is presented in this dissertation.The proposed algorithm,with low-complexity to obtain the adaptive weights,does not incur the additional calculations to obtain the reduced-rank matrix and yields a satisfactory broadband beamforming performance.4.Space-frequency adaptive processing is an established method to design a broadband beamformer.The uniform decomposition method(UDM)is a common approach to design sub-band adaptive beamformer(SAB)that splits the received signal into a number of uniform sub-bands.However,the UDM has redundancies in decomposed sub-bands at high frequencies over the passband.In this dissertation,we propose a number of techniques to overcome this issue.By proposing a novel relative bandwidth method(RBM),we ensure that the relative bandwidth of each sub-band is the same.Using this as a basis,we present a non-uniform decomposition method(NUDM)such that the NUDM has fewer sub-bands than the conventional UDM,leading to reduced computational complexity.We then extend NUDM method to provide a fast variant of the non-uniform decomposition SAB(FNUD-SAB).We ensure that the sub-band frequencies and corresponding adaptive weights are available as part of the proposed FNUD-SAB method.With undistorted response to the desired signal and effective anti-jamming capability,the new beamformer reduces the computational complexity by reducing the number of sub-bands.Simulation results highlight the effectiveness of the proposed methods.