Jitter compensation in sampling via polynomial least squares estimation

Sampling error due to jitter, or noise in the sample times, affects the precision of analog-to-digital converters in a significant, nonlinear fashion. In this paper, a polynomial least squares (PLS) estimator is derived for an observation model incorporating both independent jitter and additive noise, as an alternative to the linear least squares (LLS) estimator. After deriving this estimator, its implementation is discussed, and it is simulated using Matlab. In simulations, the PLS estimator is shown to improve the mean squared error performance by up to 30 percent versus the optimal linear estimator.