Method or Apparatus Equalize Highly Nonlinear Communication Channels using Deep Neural Networks

This idea leverages artificial intelligence, specifically deep neural networks (DNNs) at the receiver to equalize signals transmitted through communication channels dominated by nonlinear distortion (NLD). DNN may enable superior receiver equalization performance over alternative mitigation techniques including transmitter pre-distortion, power amplifier output-power-backoff (OBO), and Bayesian signal detection. Receiver's equipped with NLD equalization may be operated more efficiently, enabling better network performance and economics.


  • Less complex than Bayesian signal estimation method alternative.
  • Cost reduction opportunities in many of today's receiver implementations Improved network robustness against NLD.


Binghamton University RB546

Patent Information:
Technology/Start-up ID: