Algorithmic extensions of Su-Wong-Ho linear MMSE estimator for large-magnitude Levy-process phase-noise

A new linear minimum-mean-square error (LMMSE) estimator has recently been proposed by Su, Wong and Ho to estimate phase-noise of (possibly) large magnitude, temporal non-stationarity, and Levy distribution (which includes the Wiener distribution as a special case). This estimator can handle many different degrees of latency. The estimator is adjustable to any number of taps, which may be pre-computed offline, based on only the signal-to-(additive)-noise ratio and the phase-noise's characteristic function. This pre-computation requires no matrix inversion. The above estimator is algorithmically extended for more flexibility in latency, to select the optimum estimator-tap support-window from a wider data-observation window, and to handle newly arrived samples in a computationally efficient manner.