Deep Reinforcement Learning Based Active Queue Management for IoT Networks

Internet of Things (IoT) finds its applications in home, city and industrial settings. Current network is transition to adopt fog/edge architecture for providing the capacity IoT. However, order deal with enormous amount traffic generated by IoT devices reduce queuing delay, novel self-learning management algorithms are required at nodes. Active Queue Management (AQM) a known intelligent packet dropping techique differential QoS. In this paper, we propose new AQM scheme based on Deep Reinforcement Learning (DRL) technique introduce scaling factor our reward function achieve trade-off between delay throughput. We choose Q-Network (DQN) as baseline scheme, compare approach various schemes deploying them interface node. simulated configuring different bandwidth round trip time (RTT) values. The simulation results show that outperforms other terms jitter while maintaining above-average throughput, also verifies DRL efficient managing congestion.