QQ客服 官方微博 反馈留言

	基于缓冲区驱动的自适应视频多播系统
	刘晓颖 ; 刘晓冬 ; 谭鹏 ; 林孝康 ; Xiaoying Liu ; Xiaodong Liu ; Peng Tan ; Xiaokang Lin
	2010-07-15 ; 2010-07-15
会议名称	第一届建立和谐人机环境联合学术会议（HHME2005）论文集 ; 第一届建立和谐人机环境联合学术会议（HHME2005） ; 中国昆明 ; CNKI ; 中国计算机学会、中国图象图形学学会、ACM SIGCHI中国分会、清华大学计算机科学与技术系
关键词	自适应传输 接收方缓冲区驱动 实时视频多播 adaptive transmission receiver-buffer-driven real-time video multicast TN919.8
其他题名	Buffer-driven System for Adaptive Video Multicast
中文摘要	视频多播已经成为IP多播领域的重要研究课题。为了解决现有网络基础上的实时视频多播问题,本文利用基于控制理论的自适应传输机制,提出了一种接收方缓冲区驱动的自适应视频多播系统,实现了在不可预约带宽网络环境中的视频自适应传输。仿真结果验证了方法的有效性,而且可适用于不同的拥塞控制算法和编码算法,扩展性良好。; Video multicast becomes one of the major fields in IP multicast. It is highly desirable to meet the real-time requirement of video multicast on the Internet. In this paper, a receiver-buffer-driven system is proposed for adaptive video multicast. The proposed system adapts an adaptive transmission scheme that is based on control theory. The frame of this transmission mechanism is as: the video source is send into the video stream of sender buffer area via the adaptive streaming multimedia module, and then is launched into the client by the transmission sub-system. Where, the adaptive streaming multimedia module mainly includes four components: the monitor of sender buffer area, the observer of receiver buffer area, the controller of video quality, the scheduler of data frames and the shaper of rate. The monitor of sender buffer area is used for watching the sender buffer occupancy Bs(t). The observer of receiver buffer area is applied to calculate the receiver buffer occupancy Br(t). According to the output of receiver buffer observer, video quality controller calculates a reference video quality Qref(t) is used by rate shaper to scale video quality. In our scheme, we use PI (Proportion and Integral) controller from control theory as video quality controller. In order to prevent sender buffer underflow and overflow, we use a sender buffer driven frame scheduler approach. That is, once the sender buffer occupancy is below a given threshold, frame scheduler will fetch from the storage device the following frame. Then, the fetched frame will be processed by rate shaper and put into sender buffer. The frame scheduler will repeat this process, stop this process when sender buffer occupancy is above the threshold, and restart this process when the sender buffer occupancy goes below the given threshold again. For receiver buffer occupancy reference Brref, too large Brref may cause large start-up delay or poor presentation video quality at the beginning of video playing. However, too small Brref may limit the ability of video server to smooth presentation video quality during time of congestion. So it is suitable to choose a moderate Brref based on the receiver buffer capacity and the amount of data preloaded to receiver buffer before playing video. For Ki and Kp, on one hand, relatively large Ki and Kp, will avoid receiver buffer underflow and overflow. On the other hand, the smaller Ki and Kp the smoother presentation video quality. Thus, in order to prevent receiver buffer underflow or overflow while achieving smooth video presentation quality, it is suitable to dynamically set parameters as: if receiver buffer is near underflow or overflow, choose large Ki and Kp to prevent receiver buffer underflow or overflow, else, controller switches to smaller Ki and Kp, to achieve more smooth playback quality. Though congestion control mechanism is able to minimize, packet loss ratio and avoid excessive packet delay, we have studied how to accommodate network delay and packet loss for this. Simulation results validate the effectiveness of the proposed scheme. This system can be used for different congestion control algorithms and source encoding techniques. At the same time, the proposed system realizes adaptive video transmission under various Internet conditions.; 国家重点自然科学基金(编号:60432030)
语种	中文 ; 中文
内容类型	会议论文
源URL	[http://hdl.handle.net/123456789/69056]
专题	清华大学
推荐引用方式 GB/T 7714	刘晓颖,刘晓冬,谭鹏,等. 基于缓冲区驱动的自适应视频多播系统[C]. 见:第一届建立和谐人机环境联合学术会议（HHME2005）论文集, 第一届建立和谐人机环境联合学术会议（HHME2005）, 中国昆明, CNKI, 中国计算机学会、中国图象图形学学会、ACM SIGCHI中国分会、清华大学计算机科学与技术系.

个性服务

查看访问统计

相关权益政策

暂无数据

收藏/分享

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。