Orthonormal lattice filter—a multistage multichannel estimation technique

摘要: The orthonormal lattice filter algorithm is a unified recursive technique to estimate the least‐squares coefficients of broad class linear models. It consists sequence two recursions. First, Gram‐Schmidt orthonormalization procedure estimates set inner products from multiple signal vectors, for example, multichannel sampled observations. A type network structure, suitable adaptive applications, implements procedure. Second, recursion introduced in this paper transforms into model coefficients. In contrast with conventional algorithms, does not require covariance matrix design filters but directly We consider number special cases model, example rational polynomial autoregressive and autoregressive‐moving average known unknown inputs. addition, we introduce autoregressive‐orthonormal inpu...