Multi-objective Adaptive Routing in Identification Self-organized Network Based on Reinforcement Learning

The Identification self-organized Network is a newly developed private self-organized network that utilizes identification network technology and imposes stricter requirements on differentiated quality of service guarantees. However, due to the highly dynamic changes of the network and the limited use of network resources, routing design for self-organized network presents significant challenges. The existing wireless self-organized network routing protocols are difficult to meet the requirements of the new private identification self-organized network for differentiated service quality assurance, as well as the requirements for low delay and low energy consumption. This paper proposes a new multi-objective adaptive routing protocol based on reinforcement learning for identification self-organized networks. The proposed protocol provides differentiated quality of service guarantees with low delay and low energy consumption for different link states. It adapts to dynamic network topology and resource constraints by designing a differentiated reward function and adaptively adjusting Q-learning parameters. The simulation results show that this method achieves lower delay, higher packet delivery rate, and lower energy consumption while ensuring differentiated service quality, as compared to the active routing DSDV protocol and on-demand routing AODV protocol.