In vehicular ad-hoc networks (VANETs), store-carry-forward approach may be used for data sharing, where moving vehicles carry and exchange data when they go by each other. In this approach, storage resource in a vehic...In vehicular ad-hoc networks (VANETs), store-carry-forward approach may be used for data sharing, where moving vehicles carry and exchange data when they go by each other. In this approach, storage resource in a vehicle is generally limited. Therefore, attributes of data that have to be stored in vehicles are an important factor in order to efficiently distribute desired data. In VANETs, there are different types of data which depend on the time and location. Such kind of data cannot be deployed adequately to the requesting vehicles only by popularity-based rule. In this paper, we propose a data distribution method that takes into account the effective life and area in addition to popularity of data. Our extensive simulation results demonstrate drastic improvements on acquisition performance of the time and area specific data.展开更多
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...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。展开更多
文摘In vehicular ad-hoc networks (VANETs), store-carry-forward approach may be used for data sharing, where moving vehicles carry and exchange data when they go by each other. In this approach, storage resource in a vehicle is generally limited. Therefore, attributes of data that have to be stored in vehicles are an important factor in order to efficiently distribute desired data. In VANETs, there are different types of data which depend on the time and location. Such kind of data cannot be deployed adequately to the requesting vehicles only by popularity-based rule. In this paper, we propose a data distribution method that takes into account the effective life and area in addition to popularity of data. Our extensive simulation results demonstrate drastic improvements on acquisition performance of the time and area specific data.
文摘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。
