An improved bearing estimation algorithm for randomly perturbed arrays without calibration

Calibration-free algorithms which perform bearing estimation on the basis of a Toeplitz covariance matrix reconstructed from the observed matrix were proposed by Y.M. Chen, J.H. Lee and C.C. Yeh (Proc. ICASSP p.2917-20, 1988) for a linearly periodic array whose sensor locations are randomly perturbed. In the present work, the methods of Chen et al. (1988) are improved upon by utilizing an iterative algorithm to reconstruct from the observed matrix a covariance matrix satisfying both Toeplitz and eigenstructure properties. Computer simulations show that the iterative method is superior to the method of Chen et al. (1988) in terms of the cost of computations.<>