FedDQN-TSR

Federated Deep Q-Learning for Traffic Sign Recognition in IoV

FedDQN-TSR applies Federated Deep Q-Network learning to resource-aware traffic sign recognition in Internet of Vehicles (IoV) environments.

Key Contributions

  • Federated RL framework where vehicles collaboratively train a DQN without sharing raw sensor data
  • Resource-aware scheduling that adapts to vehicle compute and bandwidth heterogeneity
  • Demonstrates convergence and accuracy improvements over centralized and naive FL baselines
  • Published at ICECTE 2026 (IEEE)

References