A fast algorithm for unmodified ARMA spectrum estimation

A recursive algorithm for finding the parameters of Cadzow\´s "high performance" ARMA (Autoregressive and Moving Average) spectral estimation model is developed. The method does not require any data modification techniques which have been often used to develop computationally fast algorithms. The development of the algorithm is based on the shift invariant structure of the covariance matrix and orthogonal projection operators. The number of arithmetic operation at each arrival of new data points have been reduced to the order of p with p being the order of denominator coefficients. In the paper, development of fast algorithm and its numerical examples of ARMA model spectral estimation are described.