Dynamically Programmable Non-uniform THz Surfaces and Analog Neural-Network Accelerators through Inverse Design

Abstract

Spatio-temporal control, synthesis and detection of electromagnetic fields at deep sub-wavelength scales that is dynamically reconfigurable at THz frequencies can enable transformative changes in the field enabling a wide array of new applications across sensing, imaging and communication. In this proposal, we present basic research approaches to realize programmable THz surfaces on chip through inverse design methodologies that allow rapid manipulation of THz fields across frequency, angle of incidence and polarization at GHz speed. We will also show how such programmable surfaces can allow linear computation in the electromagnetic domain, and can act as rapid and efficient analog neural network accelerators for real-time decision making capabilities at the edge.

Document Details

Document Type
DoD Grant Award
Publication Date
Feb 29, 2024
Source ID
FA95502310176

Entities

People

  • Kaushik Sengupta

Organizations

  • Air Force Office of Scientific Research
  • Trustees of Princeton University
  • United States Air Force

Tags

Fields of Study

  • Physics

Readers

  • Electromagnetic Wave Scattering and Antenna Radiation Engineering
  • Neural Network Machine Learning.
  • Robotics and Automation.

Technology Areas

  • AI & ML
  • AI & ML - Machine Learning Algorithms
  • AI & ML - Neural Networks