Fast and Precise High-Speed Channel Modeling and Optimization Technique Based on Machine Learning

This letter proposes a fast and precise high-speed channel modeling and optimization technique based on machine learning algorithms. Resistance, inductance, conductance, and capacitance (RLGC) matrices of a high-speed channel are precisely modeled by design-of-experiment method and artificial neural network. In addition, an optimal channel design, which achieves minimum channel loss and crosstalk, is investigated within short time by a genetic algorithm. The performance of the proposed technique is validated by simulations up to 20 GHz.

I. Introduction

REQUIREMENT of bandwidth of electrical digital systems in modern data communications is continuously increasing. With the substantial increase in the demand for higher performance and higher data rate, the speed of modern digital electronic systems is limited more by the I/O channels, because the bandwidth of the I/O channel cannot keep up with the increasing speed of the processing units. Channel design to maximize bandwidth becomes critical. Thus, validation of designed high-speed channel to guarantee the quality of transmitted digital signal through the channel is always necessary.