Research On DOA Estimation And The Ambiguities For Sparse Arrays

The estimation of direction of arrivals (DOA) is an important research field ofarray processing. Compared with the traditional arrays, the sparse arrays which have thesame number sensors as the filled arrays have bigger antenna aperture and provide abetter performance. However, they are subject to manifold ambiguity.Then novel methods and processes are explored and investigated for thoseproblems. The main results are as follows:1. A DOA estimation method by using virtual array is presented in this paper. Onthe one hand, the output of the old array is calculated according to some rules to get thecovariance matrix of a new array named "virtual array". On the other hand, themanifolds of the old array and the virtual array are calculated in a set of DOA, and thecovariance matrix of the virtual array is calculated by the virtual array manifold and thesignal covariance matrix estimated by the old array. Two vectors are constructed by thetwo covariance matrix and the dot product of the two vectors is considered by genealgorithm. The simulations show that the method works well whether the signals arecorrelated or not for any array structure.2. An improved DOA estimation method by using virtual array is presented inthis paper. The improved method changes the one multiple-dimension globaloptimization of the basic method into several1-D global optimizations, so thecomputation load is reduced significantly. If the results estimated by the virtual arraymethod are the initial values of the Newton-like algorithm and the cost function iscalculated by the Newton-like algorithm, the accuracy will be improved.3. A method based on the Multiple Signal Classification (MUSIC) algorithm tosolve the manifold ambiguity for sparse array is proposed in this paper. The methodconsists of two steps. The first step is to obtain all the DOAs, including true andspurious DOAs, by traditional MUSIC. The second step is to estimate the power valuesof the all DOAs by substituting the all DOAs to a cost function. And the Newton-like orthe particle swarm optimization (PSO) method is used to estimate the power values. The power values of spurious DOAs are very small or tend to zero compared with the valuesof the true DOAs. The true DOAs are then differentiated easily from the spurious DOAswith the power values. Simulation results demonstrate the effectiveness and thefeasibility of our method.4. A method based on the MUSIC algorithm using substrate is proposed. Theambiguity is caused by the fact that the steering vector of spurious DOA of non-trivialambiguities is a linear combination of the steering vectors of true DOAs. When somesubstrates, whose parameters are different from each other, are added at some front-endsof the elements of an array, the equivalent positions of the elements change for differentDOAs after adding the substrates on the array, and the spurious peaks will disappear ornot overlap on the old peaks. Two types' substrates, the planar substrates andsemi-sphere substrates, are researched. Simulation results demonstrate the effectivenessand the feasibility of the method. Meanwhile the simulation results also show that theproposed method can resolve the trivial ambiguities even if the true DOA and thespurious DOA are confined to0,180.5. The ambiguity of the sparse array that some substrates are added at thefront-ends of the elements is studied. It is demonstrated that the sparse array with onesubstrates adding at the front-ends of the elements has no ambiguities.The DOA estimation algorithm presented in this paper can be applied to any arraystructure that is uniform or non-uniform. It works well even if the signals are correlated.The algorithms of resolving manifold ambiguities for sparse array have betterperformance. The conclusion that the sparse array with one substrates adding at thefront-ends of the elements has no ambiguities provides an efficient way to design asparse array which has no ambiguities.