Abstract
The IBIS algorithmic modeling interface (IBIS-AMI) has become the standard methodology to model Serializer/Deserializer (SerDes) behavior for end-to-end high-speed serial link simulations. Meanwhile, machine learning (ML) techniques can mimic a black-box system behavior. This article proposes the self-evolution cascade deep learning (SCDL) model to show a parallel approach to effectively modeling adaptive SerDes behavior. Specifically, the proposed self-guide learning methodology uses its own failure experiences to optimize its future solution search according to the prediction of the receiver equalization adaptation trend. The proposed SCDL model can provide the high-correlation adaptation results, while the adaptation simulation time is much faster than conventional IBIS-AMI models.
Highlights
W ITH the receiver adaptation algorithm, a robust serial link can be built regardless of the transceiver setting and channel characteristics
The machine learning (ML)-based technique can speedup the adaptation process compared with the conventional IBIS-AMI modeling approach
The self-evolution cascade deep learning (SCDL) model figures out that using the pseudorandom binary sequence (PRBS) data will significantly increase the model training time and model complexity and produce low-prediction accuracies for all the codes no matter which PRBS data pattern is used, in the range of 20%–30%
Summary
W ITH the receiver adaptation algorithm, a robust serial link can be built regardless of the transceiver setting and channel characteristics. In the conventional IBIS-AMI model development, the knowledge to the underlying circuit is required to build an efficient model and the details of the adaptation algorithm. To decouple the need of circuit knowledge in the model design, all the data are the receiver input waveform and the receiver adaptation output codes. With this limited information, the problem fits itself as a black-box problem: a system viewed in terms of its inputs and outputs, without any knowledge of its internal workings. To the best of our knowledge, there is no research focus on leveraging ML models on receiver CTLE and DFE adaptation behavior To tackle these challenges, an ML modeling framework is built to eliminate the time-consuming model tuning step in the ML model. The ML-based technique can speedup the adaptation process compared with the conventional IBIS-AMI modeling approach
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
More From: IEEE Transactions on Components, Packaging and Manufacturing Technology
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.