Givens rotation based multichannel least squares adaptive lattice algorithm

By combining the spatial lattice filter with a single channel hybrid lattice/QR least squares (LS) algorithm, a multichannel LS lattice algorithm based completely on QR decomposition of the multichannel data matrix directly is presented in the paper. The parameter vector to be estimated in each channel can be obtained directly from internal variables in the present algorithm without a backsolve operation. The parameter identification problem to a multichannel autoregressive model is solved.<>