Study On Direction Of Arrival Estimation Algorithms For Correlated Signals In Smart Antenna

In recent years, there has been a growing interest in developing high-resolution directionof arrival (DOA) estimation techniques in smart antenna. Many algorithms, such as MUSIC,ESPRIT, have been developed over the years and provide satisfactory performance. However,in the environment of mobile communication, those high-resolution methods may fail to workwhile the signal transmits through multi-paths and becomes correlated signals for gettingdifferent time–delay and loss. Nowadays researches on the decorrelation algorithm areincreasing sharply. The decorrelation algorithm for DOA estimation of correlated signals ismainly researched in this paper. The details are illustrated as follows:Firstly, an improved Unitary ESPRIT algorithm is proposed for DOA estimation ofcorrelated signals. This algorithm uses both row elements and corresponding column elementsof array covariance matrix to construct two Toeplitz matrixes, and a new covariance matrix isobtained by taking average of these two matrixes. This process is equivalent to theforward-backward smoothing technology and reduces the influence of the array error andnoise. The new covariance matrix is changed into a real matrix by unitary transformation, sothat the calculation quantity of this algorithm is reduced to 75% of the ESPRIT-LIKEalgorithm. The simulation results demonstrate the effectiveness and efficiency of thisalgorithm.Secondly, a new decorrelation method based on 2q-MUSIC algorithm is presented forDOA estimation of correlated signals. To achieve DOA estimation in coherent multi-pathenvironment, the algorithm constructs a 2q-order cumulates matrix with 2q-MUSIC and thensmooth it with the spatial smoothing technique for the full rank of the matrix. Simulationresults show that the algorithm in coherent multi-path environment not only accuratelyestimates the direction of the arrival, handles more sources and improves the accuracy ofestimation, but also gets the same result in any additive Gaussian noise environment.Lastly, an eigenvector reconfiguration algorithm based on spatial differencing is proposedfor DOA estimation of correlated signals while uncorrelated and correlated signals coexist.The uncorrelated signals are first estimated using conventional subspace methods. Then theireffects are eliminated by constructing a spatial differencing matrix which includes onlycorrelated signals. And then, with the eigenvectors of the spatial differencing matrix, theeigenvectors of correlated signals that belong to the multi-path of different sources are used toconstruct new Toeplitz matrixes to estimate correlated signals. The number of signals resolvedby improved algorithm can exceed the number of array elements. The simulation resultsdemonstrate the effectiveness and efficiency of this algorithm, especially at low SNRscenarios.