Deep Reinforcement Learning Based Algorithm for Symbiotic Radio IoT Throughput Optimization in 6G Network

Internet of Things (IoT) -based 6G is expected to revolutionize our world. Various candidate technologies have been proposed to meet IoT system requirements based on 6G, symbiotic radio (SR) is one of these technologies. This paper aims to use symbiotic radio technology to support the passive Internet of things and enhance uplink transmission performance. The IoT tag information is sent to the cloud for analysis through a macro base station (MBS) or a wireless access point (WAP), where the smartphones are used as a relay to transmit this information to the MBS or WAP. In this paper, an optimization problem was formulated into two phases to maximize the total throughput of the system. The first phase is, the problem of achieving the optimum mode selection of the LTE or Wi-Fi Network, aiming to maximize the system throughput. A matching game algorithm is used to solve this problem. Second phase, the problem of achieving optimum clustering of tags, where the tags are divided into virtual clusters, and finding which smartphones' LTE/Wi-Fi downlink signals all cluster members can ride to maximize the system throughput. A double deep Q-network (DDQN) model was proposed to solve this problem. Simulation results show that our proposed algorithms increase the total system data rate by an average of 90% above the system using the LTE network first without DDQL algorithm. Furthermore, it enhances the capacity of the system on average by 100% above LTE network first system without the DDQL algorithm.