Efficient sensitivity aware assessment of high-speed links using PCE and implications for COM

In this paper, the Polynomial Chaos Expansion (PCE) with a Stochastic Testing (ST) sampling scheme is demonstrated to be a highly efficient method for quantifying stochastic design variations. The results of PCE are correlated with those of the established approaches Monte Carlo as well as Response Surface Method based on Design of Experiments.

Sensitivity analysis of a high-speed interconnect

The PCE workflow proposed in this paper can increase the efficiency and effectiveness of link design and validation. It extends compliance checks with additional information on the significance of design parameters and quantifies the consequences of their uncertainty. The paper goes into detail on the following topics:

• Introduction to commonly used performance metrics and methods of uncertainty quantification in high-speed link design with a particular focus on PCE and ST
• The main two aspects of the proposed modeling framework: first, extending an open-source PCE library by ST, and second, the set up of a sensitivity-aware channel model
• Explore the similarities of PCE and RSM to propose a compound, computationally efficient, and self-validating surrogate model.

This paper was co-written by authors from Hamburg University of Technology, Cisco Systems, and Siemens.

To learn more about SERDES, visit.