Q-learning based Backoff Time Selection Mechanism in IEEE 802.15.4 Unslotted CSMA/CA

Key Points

• The aim is to optimize backoff time selection in IEEE 802.15.4 Unslotted CSMA/CA using Q-learning to enhance network performance.
• Used Q-learning for backoff time selection instead of random selection.
• Evaluated performance using Network Simulator 3 as the node count increased.
• Assessed throughput, delay, and network overhead, including scenarios with hidden nodes.
• Showed a 42.8% improvement in throughput compared to traditional CSMA/CA with the same backoff time.
• Significantly enhanced performance even in the presence of hidden nodes.