带宽约束下多客户端视频流体验质量动态优化  

Dynamic optimization of multi-client video streaming QoE under bandwidth constraints

作　　者：谭光 丁佳龙 TAN Guang;DING Jia-long(School of Intelligent Systems Engineering,Shenzhen Campus of Sun Yat-sen University,Shenzhen 518107,China;School of Intelligent Systems Engineering,Sun Yat-sen University,Guangzhou 510006,China)

机构地区：[1]中山大学·深圳智能工程学院,广东深圳518107 [2]中山大学智能工程学院,广东广州510006

出　　处：《计算机工程与设计》2023年第7期1921-1927,共7页Computer Engineering and Design

基　　金：国家自然科学基金项目(61772509);广东省自然科学基金项目(2019A1515011066)。

摘　　要：为解决多客户端的带宽资源分配问题,提高用户体验质量(quality of experience,QoE),建立多客户端视频流的体验质量优化框架。针对已有视频流算法在多客户端领域的缺陷,基于模型预测控制算法提出一个多客户端带宽动态调度算法,根据每个客户端的带宽预测情况对它们进行动态资源分配,通过提高带宽利用率进而提升总体用户QoE。在HSDPA网络带宽轨迹上的仿真结果表明,相比各客户端平均带宽分配方式,优化方法在总体用户体验质量上提升42.6%以上,相比最新的Minerva方案提升了7.8%。To solve the problem of bandwidth resource allocation of multiple clients and improve the quality of experience(QoE)of users,a framework for optimizing the quality of experience of multi-client video streaming was established.Aiming at the shortcomings of existing video streaming algorithms in the field of multi-clients,a multi-client bandwidth dynamic scheduling algorithm was proposed based on the model predictive control(MPC)algorithm.Resources were dynamically allocated to the clients,and the overall user QoE was improved by improving bandwidth utilization.Simulation results on the HSDPA network bandwidth traces show that the total QoE of the optimized method is improved by at least 42.6%compared to the average bandwidth allocation method.It outperforms the state-of-art approach,Minerva,with an increase of 7.8%.

关 键 词：视频流 体验质量 多客户端 自适应比特率 带宽预测 决策算法 资源调度 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]