1989 Volume 2 Issue 3 Pages 88-96
In this paper, applications of the circular lattice (CL) method for multichannel signal processing to ARMA modeling are presented. After a review of the CL method, the covariance characterization of an ARMA process is discussed by embedding an ARMA process into a degenerate two-channel AR process. The connections with the method of Mullis-Roberts and the theory of Markov-cover by Anderson and Skelton are pointed out.
Then, a new ARMA lattice filter is proposed and the effect of coefficient quantization on the frequency response of our lattice filter is compared with those of other ARMA lattice filters. Finally, an ARMA parameter estimation problem based on the impulse response data is treated by using the Burg type CL algorithm. Some numerical results are presented to show the effectiveness of the proposed method.