中文摘要：

随着互联网技术的快速发展以及智能设备的普及,基于HTTP的动态自适应流媒体（Dynamic Adaptive Streaming over HTTP,DASH）业务发展迅速.但在带宽受限网络中,大规模用户的视频请求,将会加重网络负载,严重影响网络带宽资源的有效利用,同时用户码率调节缺乏全局协调控制机制,容易造成网络拥塞.针对软件定义网络中的DASH视频传输业务,将视频业务提供商长期平均收益最大化作为优化目标,设计并实现了基于神经元动态规划的DASH视频路由和用户码率调节联合决策算法.最后,通过在Mininet平台上建立SDN（Software-Defined Networking）网络环境并进行对比实验,我们验证了本文提出的联合决策算法能够提高网络带宽资源利用率,最大化DASH视频业务提供商长期平均收益.

英文摘要：

With the rapid development of Internet technology and the popularity of smart devices, DASH （ Dynamic adaptive streaming over Hypertext Transfer Protocol）has been used widely. However, with limited bandwidth resources, a large number of DASH video requests will increase the network load and seriously affect the rational use of network resources, moreover,the DASH user＇s bitrate adjustment lack of global coordination and control mechanism, which will easily cause network congestion. In this paper, we design and implement a DASH video transmission routing and bitrate control algorithm based on the neuro-dynamic programming to max the long-term average rewards for video service provider. In the end, we establish a simulation environment through mininet and compare the experiment result of our algorithm with the OSPF algorithm, which indicates our algorithm has more excellent performance.