Deep Learning for Tunable Metasurfaces

Abstract

Deep Learning for Tunable MetasurfacesProject Summary/Abstract - (Approved for Public Release)Hualiang Zhang, University of Massachusetts LowellThe overarching goal of this project is to pioneer technologies to create universal deep learning framework for achieving tunable metasurfaces with large dynamic tuning range, low-profile, wide range of frequency coverage, compact size, as well as high-performance. The proposed project is realized through the fusion of electromagnetics and machine-learning theories. It is achieved by exploring deep learning techniques with the capability to predict and inversely design non-intuitive metasurfaces with tunable phase and amplitude responses at desiredfrequency bands, as well as discovering underlying physical insights. It is expected that the proposed research will revolutionize current metasurface design and lead to next-generation high-performance electromagnetic systems for various military applications.The proposed project brings several unique benefits: 1) reduction of time-consuming electromagnetic simulations for validating the performance of tunable metasurfaces; 2) finding non-intuitive device designs based on pre-determined electromagnetic response requirements for tunable metasurface designs; 3) the resulting deep learning platform is remarkably fast, accurateand demands minimal human intervention and efforts during the design process; and 4) the proposed deep learning approach can be generally applied to the design of other complex electromagnetic problems such as multi-functional electromagnetic components.If successful, the proposed research will lead to new paradigms for electromagnetic device design, which is enabled by the proposed deep-learning framework. It will also provide new tunable metasurfaces for manipulating electromagnetic signals (for applications such as beamsteering systems, tunable and multifunctional radiating apertures), which are scalable fromRF/microwave to higher frequencies. Moreover, the fundamental design approach and devices developed in this project can be applied to other electromagnetic systems. Overall, the successful implementation of the proposed effort will strengthen the electromagnetic component technologies needed for future military systems, and contribute to DoD~s capabilities and competitiveness in pioneering technology for key areas such as radar, sensing, communication, and navigation.

Document Details

Document Type
DoD Grant Award
Publication Date
Dec 16, 2019
Source ID
N000142012062

Entities

People

  • Hualiang Zhang

Organizations

  • Office of Naval Research
  • United States Navy
  • University of Massachusetts

Tags

Readers

  • Distributed Systems and Data Platform Development
  • Quantum Dot Semiconductor Device Photonics and Graphene Optoelectronic Materials and THz Physics.
  • Research Science/Academic Research

Technology Areas

  • AI & ML
  • AI & ML - DoD AI Strategy
  • AI & ML - Neural Networks