AI‑Driven Decoding of Graph‑Based Channel Codes via Reinforcement Learning | New Jersey Institute of Technology

AI‑Driven Decoding of Graph‑Based Channel Codes via Reinforcement Learning

AI‑Driven Decoding of Graph‑Based Channel Codes via Reinforcement Learning (Systems and Methods for Decoding of Graph‑Based Channel Codes via Reinforcement Learning, Tech ID: 21‑021)

Technology Overview: This technology applies reinforcement learning (RL) to decode graph‑based channel codes used in modern digital communication systems. By learning optimal decoding strategies dynamically, the system improves error‑correction performance and adaptability compared to fixed‑algorithm decoders. This AI‑enabled approach is particularly valuable for next‑generation wireless and data communication systems operating under variable channel conditions.

Industry Pain Point: Conventional decoding algorithms struggle to adapt to dynamic channel environments.

NJIT Solution: Reinforcement learning enables adaptive, high‑performance decoding across varying conditions.

Key Features & Advantages

  • AI‑driven adaptive decoding
  • Improved error‑correction performance
  • Applicable to modern channel codes
  • Supports next‑generation communication systems

Development Stage: TRL 3–4 – Algorithmic validation completed.

Target Markets

  • Wireless communications
  • Data transmission systems
  • Semiconductor and communications IP providers

Market Opportunity

  • Global wireless communications market (2026): >$800B
  • CAGR: ~6–7%
  • Projected market size (2035): >$1.3T

Commercial & IP Details

Inventors: Allison Beemer, Salman Habib, Joerg Kliewer

Patent Information:
Category(s):
Computing, Communications & Photonics
For Information, Contact:
Ikechukwu Nwabufo
IP Licensing & Marketing Manager
in49@njit.edu
Inventors:
Joerg Kliewer
Salman Habib
Allison Beemer
Keywords:
AI-Driven
Data Processing
Patent Issued
Reinforcement Learning (RL)
semiconductor
Wireless Communication