Systolic Algorithms for Recursive Total Least Squares Parameter Estimation and Mixed RLS/RTLS Problems

Total least squares parameter estimation is an alternative to least squares estimation though much less used in practice, partly due to the absence of eecient recursive algorithms or parallel architectures. Here it is shown how previously developed systolic algorithms/architectures for recursive least squares estimation can be used for recursive total least squares problems. Unconstrained as well as linearly constrained and`mixed RLS/RTLS' problems are considered.