A Robust Algorithm for PAM4 Eye-Diagram Analysis

We propose an approach for analyzing PAM4 (pulse amplitude modulation 4-level) eye diagrams that always provides a unique solution by making use of a K-Means algorithm in conjunction with a robust, shortest interval location estimator. Our motivation for developing this technique is to create an independent, nonproprietary method that can be used to compare and verify existing algorithms. Utilizing our approach, we calculate various metrics that have been developed for PAM4 standards, including the time midpoint of the middle eye, eye amplitudes, inner eye widths, and inner eye heights. We compare our results to those of a commercial channel simulator and obtain excellent agreement.