The normalized least-squares order-recursive lattice smoother
Signal Processing
Network numerical analysis for the smoother and the lagged joint-process estimator
The Journal of Supercomputing
Hi-index | 35.68 |
Conventional least squares order-recursive lattice (LSORL) filters use present and past data values to estimate the present value of a signal. This paper introduces LSORL smoothers which use past, present and future data for that purpose. Except for an overall delay needed for physical realization, LSORL smoothers can substantially outperform LSORL filters while retaining all the advantages of an order-recursive structure