Virtual prototyping tools for electronic design automation

Abstract

Virtual prototyping environments are emerging as a new generation of EDA tools. Such a tool will allow the designer to test interactively complex electronic systems on an enhanced-reality virtual workbench, by concurrently running multidomain (mechanical, electrical, thermal, etc.) what-if experiments. Virtual prototyping will shorten the design cycle, improve the product quality, and reduce the time to market. Advanced computational techniques are needed to reduce the execution time, especially for the field (EM and thermal) models used in these virtual prototyping environments. This article shows how neural networks could be used efficiently for 3D EM field modeling. Neural network models have much better real-time performance than classical numerical EM-field modeling methods, and this is particularly important when the field analysis is coupled with system optimization.