Abstract

AbstractIn terms of avoiding congestion, due to the rapid change of satellite network topology and network traffic, traditional satellite network routing algorithms have problems in computational efficiency and scalability. The rapid expansion of deep reinforcement learning urges a lot of research work to combine DRL with network routing. We propose a novel fully distributed DRL packet routing framework named Intelligent Packet Routing in Satellite Networks (IPR-SN). The fully distributed framework is designed to avoid problems caused by centralized training or centralized execution in IPR-SN. Because DRL is endowed with powerful experience learning ability and feature learning ability, agent learns the hidden features of the model by virtue of the deep structure of neural networks. It avoids choosing the path that aggravates congestion in packet path selection and formulates a better satellite network routing strategy. Through experimental analysis and comparison with other algorithms, the results show that this algorithm can effectively reduce average packet delivery time.KeywordsPacket routingSatellite networksDistributed deep reinforcement learning

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call