Development of Polynomial Chaos Based Surrogate Models for Channel Simulation

Improvements in high frequency signal transmission places requirements on meticulous jitter modeling and prediction techniques, which have to provide precision below 1 ps. In general timing jitter is divided into random and deterministic jitter, where the latter includes data dependent jitter, which is the focus of this paper, since it can be the dominant cause of jitter and it has proven to be challenging to model. Traditionally, jitter arising in eye-diagrams of high speed channels is modeled using a transient simulation with a length of thousands or millions of bits. This approach can be very time consuming. Hence, statistical eye analysis techniques are developed, providing for fast prediction of eye diagram and jitter [1]. However these techniques are only applicable to linear systems, which can be an important constituent for jitter in high speed channel simulation.