Q-Learning-Based Opportunistic Routing Protocol for Flow-Guided Nanonetworks

To solve the problems of the high packet loss rate and weak dynamic adaptability of nanonetworks deployed in human blood vessels, an opportunistic routing protocol based on Q-learning (ORBQ) was proposed. In the Q-learning process, the reward function was defined by considering node survival ability, adjacent node signal strength, and link quality, making the evaluation of the reward mechanism more comprehensive. At the same time, in order to provide a better foundation for exploring routing paths, Q values were initialized by periodically sending data detection packets through nano control nodes. In addition, in order to enhance the effective response of the routing protocol to the blood flow environment, an adaptive adjustment discount factor mechanism and a dynamic learning mechanism were proposed to update the Q value. On this basis, a candidate node coordination mechanism based on Q-value was designed to reduce packet redundancy and conflicts. Finally, experiments were carried out using the NS3 simulation platform. The results show that the ORBQ protocol improve network performance in terms of energy efficiency, average end-to-end delay, and data transmission success rate.