A Study Of Array Signal Processing In The Presence Of Near-Field Scattering

The technology of array signal processing is widely used in radar,communication,sonar,medical electronics,radio astronomy,and other fields.In airborne and shipborne applications,there are usually many strong-scattering components such as wings and masts near the antenna,which are near-field scatterers existing in the electromagnetic environment of the antenna.Basically,when a far-field electromagnetic wave impinges upon the array antenna,the near-field scatterers also receive it and then re-radiate it.The re-radiated wave is received by the array antenna as well,which is called the near-field scattering effect.The near-field scattering signal has two characteristics: firstly,it is correlated or even completely coherent with the direct-path signal;secondly,because the scatterer is only tens of meters even several meters away from the antenna array,the near-field scattering signal impinging upon the array antenna is in the form of spherical wave,while the direct–path signal is a plane wave.What are the effects of the near-field scattering on array signal processing? This problem has not paid enough attention and systematic investigation.For this reason,this dissertation will study array signal processing techniques in the presence of near-field scattering.Particularly,this dissertation studies robust array signal processing techniques in the presence of a near-field scattering environment and various errors just as those existing in a real system.The major accomplishments and contributions are summarized as follows:1.The array signal model in the presence of a near-field scattering environment is established.In the presence of a near-field scattering environment,the signals received by an antenna array include far-field direct-path signals and near-field scattering signals.Their composite steering vectors are determined by three factors: far-field steering vector,near-field scattering matrix,and near-field scattering coefficient vector,among which the product of the near-field scattering matrix and the near-field scattering coefficient constitutes a near-field steering vector.The presence of this near-field steering vector will severely degrade the performance of DOA estimation and beamforming in conventional array signal processing.2.The influence of the near-field scattering signal on DOA estimation performance is analyzed,and a novel super-resolution DOA estimation method by using a modified steering vector that takes the near-field environment into account is proposed.It is pointed out that the near-field scattering signal will be viewed as a "new" angularly spread signal coherent with the direct-path signal.If a traditional spectral estimation method is still used for the DOA estimation,many false and spurious peaks could appear in the spatial spectrum due to the interference from the near-field scattering signal.Moreover,if the direction of the scattering signal is close to that of the direct-path signal,the DOA estimation could be significantly biased.As an example,we modify the well-known high resolution method MUSIC by using the modified array steering vector.The simulation results show that the modified MUSIC method can suppress the influence of near-field scattering signal very well.3.The influence of the near-field scattering on adaptive beamforming is analyzed.It is found out that a traditional adaptive beamforming could form a ditch in the main lobe direction because of the near-field scattering,which leads to the suppression of the target signal.To count the influence of the near-field scattering,we propose a new adaptive beamforming method through a virtual array transformation for the near-field scattering compensation.The virtual array method tries to transform the array signal including the near-field scattering into an ideal one without the near-field scattering.The simulation results illustrated that the proposed method performs very well in the presence of a near-field environment,and even outperforms the previously proposed method by directly modifying the steering vector with higher output signal-to-interference-noise ratio,lower side-lobe level,and better comprehensive performance.4.Several key issues for the robust adaptive beamforming in the presence of a near-field scattering environment are discussed.Because of unavoidable errors in the steering angle for the far-field direct-path signal,near-field scattering matrix,and the near-field scattering coefficient vector caused by many reasons in practice such as DOA estimation error,antenna rotation,fast source movement,and etc.,the robustness is desired in a practical adaptive beamforming system.The optimal estimations of these three factors will lead to a fourth-order optimization problem,which is difficult to solve.For this reason,we temporally assume that the near-field scattering matrix is precisely known,that is,the positions of the near-field scatterers can be measured accurately.In this case,the optimization of the three unknown factors becomes the optimization of two unknown factors,namely,the steering vector for the far-field direct-path signal and the near-field scattering coefficient vector.Then,we combine these two factors into one combined vector,and the fourth-order optimization problem degrades to a quadratic constrained programming problem through a transformation,and it is a convex optimization.The core of the transformation is to divide a large hypercube uncertainty set into two small ellipsoidal uncertainties from the idea of the robust Capon beamforming,which leads to a double-uncertainty-set method.The steering vector estimated by using this method is very close to the desired true steering vector including the effect of the near-field scattering,and then the optimal estimation of the true steering vector is obtained by using the method of the uncertainty set.When the near-field scattering matrix is unknown or not accurately known,two-step iterations will be used.The first step is to fix a near-field scattering matrix,and estimate the combined vector including other two vectors,i.e.,the far-field steering vector and the near-field-scattering coefficient vector,using the algorithm as described above.The second step is to optimize the near-field scattering matrix with the estimated two vectors or the combined vector.Repeat these two-steps until a sub-optimal estimation of the true steering vector is obtained.Then the optimal true steering vector is obtained by using the uncertainty set method.The simulation results illustrated that the proposed method outperforms other robust beamforming methods in the presence of near-field scattering.