无人机和无人船组成的移动自组织网络存在通信环境恶劣和网络拓扑结构变化频繁等挑战，导致网络性能变差。针对这一问题，建立以数据为中心的命名数据网络(Named Data Networking，NDN)网络架构，在此基础上提出基于深度强化学习的智能数据转发策略。利用深度强化学习实时感知网络动态变化，优化数据转发策略，设计优先采样和双重Q网络算法，加快深度强化学习收敛速度。实验结果表明，该策略可以有效降低时延并提高兴趣包满足率。
Intelligent data forwarding strategy of Named Data Networking(NDN)for UAV and USV Ad Hoc Networks
The mobile ad hoc network composed of Unmanned Aerial Vehicle(UAV)and Unmanned Ves-sel(USV)has some challenges,such as poor communication environment and frequent changes in network topology,which lead to poor network performance.To solve this problem,a data-centric NDN network ar-chitecture is established,and based on which an intelligent data forwarding strategy based on deep reinforce-ment learning is proposed.Deep reinforcement learning is used to sense the dynamic changes of the network in real time,optimize the data forwarding strategy,and design the priority sampling and double Q-network al-gorithms to accelerate the convergence speed of deep reinforcement learning.Experiment results show that the proposed strategy can effectively reduce the delay and improve the satisfaction rate of interest packets.
Named Data NetworkingMobile Ad Hoc Networkdeep reinforcement learningforwarding strategynetwork simulation
NETL
NSTL
万方数据
