DOA estimation of desired signals by cyclic ESPRIT based on noise subspace and its performance improvement

Because of the rapid development of mobile communications and ITS (intelligent transport systems), the electromagnetic wave propagation environment is becoming complex, and there is an increasing demand for techniques to estimate the direction of arrival with high resolution and to reduce the effect of interference waves. Among these, Cyclic MUSIC and Cyclic ESPRIT, in which the direction of the arriving wave can be estimated selectively by utilizing the periodic stationary property of the arriving wave, are attracting attention. The authors are proposing Cyclic ESPRIT based on the noise subspace (NS Cyclic ESPRIT) in order to remedy the performance degradation produced by the fact that the correlation matrix of the array input used in Cyclic ESPRIT is not a Hermitian matrix. However, when the number of snapshots is finite in the estimation of arrival direction by Cyclic MUSIC or Cyclic ESPRIT, the interference wave component remaining in the correlation matrix has an adverse effect on the accuracy of direction-of-arrival estimation. Consequently, adaptive spatial smoothing processing and generalized eigenvalue decomposition are introduced into NS Cyclic ESPRIT in this study, aiming at reduction of the remaining interference component and at performance improvement in direction-of-arrival estimation. A computer simulation verifies that the performance in direction-of-arrival estimation in NS Cyclic ESPRIT is improved. © 2007 Wiley Periodicals, Inc. Electron Comm Jpn Pt 1, 90(11): 95–104, 2007; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecja.20304