A Routing Algorithm Based on Prioritized Replay Double DQN for Private Internet of Things

The Private Internet of Things (P-IoT) is a network that provides IoT services to users of private networks. Effective routing protocols are essential for P-IoT to ensure optimal network performance. This paper proposes a routing algorithm called Priority Experience Replay Double Deep-Q-network (PER-DDQN), which considers both congestion and traffic load as reward functions. PER-DDQN combines deep learning and reinforcement learning to address the challenges of traditional routing algorithms in terms of learning and environmental perception. PER-DDQN enhances the learning speed of reinforcement learning and capacity for processing high-dimensional data. The simulation results indicate that the PER-DDQN algorithm outperforms in terms of convergence speed, congestion reduction, and load balancing, compared with conventional shortest path algorithms, Q-learning algorithms, and deep Q-learning algorithms.