Research On DOA Estimation Methods For Complicated Correlated Signals

Nowadays,under the highly developed information technology,the warefare form is gradually becoming from independent compaigns among different units to all-around cooperative combat.Precise strike via precision-guided munition plays an important role,and combat by missile cluster must be the dominant combat form in future warfares.Direction of arrival(DOA)is a significant parameter for target identification and localization.Hence,DOA estimation is an integral part in precise strike.However,correlated signals are usually processed in missile cluster combat.In missile detection,it is always required to handle the coexistence of radar and jammers.Also,in communications among different missiles,it is inevitable to face multipath transmission problem.Consequently,the correlated signal scenario becomes more complicated.Thus,the research of this thesis focuses on the previous-mentioned issues that are involved in the DOA estimation under complicated correlated signal scenario.Therefore,this thesis focuses on problem modeling and seeking corresponding solutions in DOA estimation for the missile cluster combat scenario.Affected by kinds of active and objective factors,e.g.,closely distributed emitters,deliberate or unconscious interference,multipath transmission of signals,we always need to operate correlated signals.Furthermore,since the communications for information share is also required in the missile group except the detection mission,this thesis considers the following 3 issues that are always faced in DOA estimation for correlated signals.First,aiming at weak angular resolution and sparsity degradation issues in classical sparse iterative covariance-based estimation(SPICE)algorithm,this thesis proposes the MEM-based weight optimization SPICE algorithm and {r,q}-SPICE,leading to an improvement on the estimation performance.In correlated signal processing algorithms,the array aperture is lost in spatial smoothing technique.Therefore,SPICE algorithm is selected as the research object in this thesis.On one hand,SPICE is essentially a sparse reconstruction algorithm,thus limited by the restricted isometry property(RIP)criterion,leading to a shortcoming in angular resolution.The maximum entropy algorithm in beamforming is involved to estimate the energy spectrum in this thesis.Then,the inverse of the estimated energy spectrum is used as the optimized weight that is exploited in iterative solution procedure of SPICE to finish DOA estimation.This is the proposed MEM-based weight optimization SPICE algorithm,and dramatically improves the angular resolution.On the other hand,classical SPICE and most improved SPICE algorithms combine the signal and noise covariance matrix together,thus leading to an increasement in the non-zero entries in the sparse vector.Thus,the sparsity is weakened,and more spurious peaks appear in the spatial spectrum.To solve this problem,the SPICE is improved in this thesis as a novel {r,q｝-SPICE algorithm,where the signal and noise are separately considered and assigned different norms.Via selecting suitable norms,the proposed algorithm can not only keep the superiorities of classical SPICE,but also obtain a sparser optimal solution and better estimation performance.Then,aiming at the problem of too many channels in polarization sensitive array,this thesis proposes the compressive polarization sensitive array structure.Thus,the number of channels is reduced,reducing the system complexity and cost and releasing the computational burden.Specifically speaking,for the scenario that radar is companied with jammers,interference signals and radar detection signals are correlated,and the received signals cannot be correctly matched to the corresponding transmitters only by DOA parameters.Thus,polarized sensitive array is exploited in this thesis,and the interference and radar detection signal are distinguished by the polarization parameters.However,for co-centered polarization sensitive antennas,each antenna is connected to multiple circuit channels,thus leading to an increasement on the total number of channels for the whole system.To address this issue,the uniform and sparse compressive polarization sensitive array are proposed in this thesis.Since the receive signal model of the proposed structure is different with that of classical polarization sensitive array,conventional estimation algorithms is improved in this thesis,and two algorithms,i.e.,the reduced-dimensional MUSIC and group sparsity reconstruction algorithm,are proposed,where high resolution and degree of freedom are obtained in each algorithm,respectively.In addition,the Cramer-Rao bound(CRB)of the sparse compressive polarization sensitive array is derived,and the corresponding theoretical performance is analyzed.Finally,aiming at the rank deficient problem in classical matrix pencil algorithm when multiple signals impinge from same or close directions,this thesis proposes a joint time difference of arrival(TDOA)and DOA estimation algorithm based on augmented matrix pencil.In multipath transmission scenario in communications,matrix pencil algorithm have outstanding robust to correlation signals,since the covariance matrix of received signals is not computed.However,multiple multipath signals may impinge from the same direction in practice,where the signals are with same DO As and different TDOAs.At this time,traditional matrix pencil algorithm has rank-deficient problem,thus compromising the singular value decomposition and leading to failures in parameter estimations.To address this issue,the traditional matrix pencil algorithm is improved in this thesis,where a pair of augmented matrix pencils is proposed.Using the augmented matrix pencils can obtain the number of signals with same DOAs via the multiplicity of eigenvalues.Thus,the rank-deficient problem is effectively solved.Combining with the matching algorithm proposed in this thesis,TDOA and DOA are paired.Numerical simulations verified the effectiveness of the proposed algorithm.