Machine Learning Assisted Evolutionary Algorithm for Device-Circuit Multi-Objective Co-optimization of Actively Shielded MWCNT Interconnects

In this paper, a neural network assisted evolutionary algorithm is developed to optimize the signal integrity features of multiwalled carbon nanotube (MWCNT) interconnects driven/loaded by nanosheet field effect transistor (NSFET) inverters in presence of active shielding. The key attribute of the proposed evolutionary algorithm is its ability to leverage analytic expressions of the multi-objective cost functions of the interconnects learned by neural networks and apply metaheuristic approaches to identify an optimal set of values of the design parameters of the NSFET devices and the signal and shield lines.