Parallel adaptive rooting algorithm for general frequency estimation and direction finding

An adaptive rooting algorithm, is proposed for tracking the nonstationary roots of null spectra for data from sensor arrays, or time series. Without the assumption of uniformity for linear arrays, this adaptive-rooting algorithm associated with the zero-extraction technique is further extended to construct a parallel-effective-rooting processor. The adaptive algorithm tracks all the principal roots and reduces the computational complexity from order N³ to order N². Simulations show better resolution performance of the suggested algorithm than that of the usual of spectral-based high-resolution techniques such as minimum-norm or MUSIC, for both uniform and nonuniform cases. Simulations also reveal that its convergence speed competes with those of recent fast adaptive eigen-structure algorithms for tracking the nonstationary parameters.