New Reward-Clipping Mechanism in Deep -Learning Enabled Internet of Things in 6G to Improve Intelligent Transmission Scheduling

Sixth-generation (6G) networks and apps have lately benefited from the use of artificial intelligence (AI) to improve a significant amount of data. The integration of AI with 6G can help support green energy and sustainable system by overcoming the complexity of network flaws. The Internet of Things (IoT) utilizes artificial intelligence (AI) to improve the management of large amounts of data, reduce energy consumption, regulate traffic, and facilitate data storage. The primary difficulty in IoT is creating intelligent agents that can enhance smart packet transmission scheduling for Ultra Reliability Low Latency Connection (URLLC). The best channel to employ for smart packet transmission scheduling in the IoT must have a low estimate Packet Error Rate (PER), as well as minimal packet delays from channel errors, and retransmissions. To improve smart packet transmission scheduling by shortening the interval between the estimated and target action value, we propose a Generative Adversarial Network and Deep Q Network (GAN-DQN). To avoid significant critical fluctuations in the target action value, GAN-DQN training is based on reward correction to evaluate the value of each action for accurate states. The simulation results demonstrate that the proposed GAN-DQN increase IoT system reliability by reducing the packet loss caused by various multiuser arrival at a BS while cutting Transmission Delay (TD) to improve intelligent transmission scheduling and power consumption.