Showing papers by "Martin Morf published in 1984"
TL;DR: Convergence analysis results give a positive realness condition for the convergence of the MR algorithm to the true RMFD of the system.
Abstract: Multivariable identification algorithms are usually designed only for left matrix fraction description (LMFD) models. In this paper we consider recursive identification algorithms for right matrix fraction description (RMFD) models with diagonal denominator matrices. The algorithms are of prediction error (PE) and model reference (MR) type. Results from simulations illustrate the performance of the algorithms.
12 citations
TL;DR: A new decision-directed (DD) procedure is introduced to address the binary detection problem for nonstationary priors using an explicit, recursive finite-dimensional filter applied to the problem of computing the conditional expectation of the stochastic rate of the discrete-time point process representing the output of the detector.
Abstract: A decision-directed receiver uses the estimate of the rate of the discrete-time point process (DTPP) output of the (0-1) detector as the unknown prior probability of the Bernoulli input sequence. This correspondence applies the results of Segall [5] to permit the estimation of a time-varying rate.
5 citations
19 Mar 1984
TL;DR: A square root normalized version of the feedback ladder algorithm for the identification of the parameters of a moving average model and the essential feature of the FBLA, that of being able to read out directly the estimated coefficients of the -monic-polynomial model is not modified.
Abstract: We have presented a square root normalized version of the feedback ladder algorithm for the identification of the parameters of a moving average model. The number of equations needed is reduced from eight in the unnormalized case to just five. The complexity of the equations increases but the procedure is justified because it seems to lead to a more convenient hardware realization. Moreover, this realization would be completely similar (for the backward and forward residuals lines) to the CORDIC processors implementation already proposed for the feedlorward ladder algorithms (FFLA). A possible disadvantage is that three of the variables used may have magnitudes greater than one. However the essential feature of the FBLA, that of being able to read out directly the estimated coefficients of the -monic-polynomial model is not modified.
3 citations