Scalable Analog Neural networks (ScAN)

Abstract

Building upon technologies discovered under the Fast Event-based Neuromorphic Camera and Electronics (FENCE) program, the Scalable Analog Neural networks (ScAN) program will increase neural network (NN) inferencing capabilities at the edge and simultaneously reduce the size, weight, and power (SWaP) needs of edge platforms. Currently, sensor outputs are digitized at the edge, which consumes SWaP and limits capabilities of edge platforms, but are then transmitted for processing at the command center. ScAN aims to skip or delay the digitization step and implement analog inferencing and compression techniques directly on the analog sensor data at the edge. ScAN objectives are to enable 2000-fold reduction in SWaP for processing of sensor data. ScAN will enable intelligence generation at the edge for missions that collect large amounts of sensor data, such as hyper-spectral imaging for unmanned aerial systems.

Document Details

Document Type
Accomplishment
Publication Date
Oct 01, 2025
Source ID
051ea5d1e3f15de449e420854670b54e

Tags

Readers

  • Computer Science/Computer Engineering/Data Science/Digital Signal Processing.
  • Computer Vision.
  • Distributed Systems and Data Platform Development

Technology Areas

  • AI & ML
  • AI & ML - Neural Networks
  • Autonomy
  • Microelectronics

Related Documents