Title :
Is P2P Assisted Video Accelerator Feasible and Efficient?
Author :
Chen, Yishuai ; Chen, Changjia ; Guo, Dan
Author_Institution :
Sch. of Electr. & Inf. Eng., Beijing Jiaotong Univ., Beijing, China
Abstract :
As the penetration of broadband Internet access, the online Internet video is prevailing. The peer-to-peer (P2P) assisted video accelerator (VA) attracted much attention recently. Based on the real-world measurement results of online videos and users´ viewing behavior, we prove the P2P VA method is feasible and efficient. We first prove, although the distribution of clicks on online videos is long-tail and only a few videos have high click rate, these videos account for considerable views, which means considerable clicks occur when multiple users are online simultaneously, and therefore the P2P method is feasible. We second prove its efficiency by using a large-scale real-world user click trace to do trace-driven simulation. The simulation results show, in the worst case in which users shut down the system after they finishes the view, the method can still decrease the server load by 25.6%. When the user is cooperative and keeps the system up for additional 30 min after he/she watches the video, the server load can be decreased by 50.6%.
Keywords :
behavioural sciences computing; media streaming; peer-to-peer computing; broadband Internet access; large-scale real-world user click trace; online Internet video; peer-to-peer assisted video accelerator; trace-driven simulation; users viewing behavior; Acceleration; Internet; Large-scale systems; Length measurement; Peer to peer computing; Portals; Streaming media; Video sharing; Watches; YouTube;
Conference_Titel :
Computer Network and Multimedia Technology, 2009. CNMT 2009. International Symposium on
Conference_Location :
Wuhan
Print_ISBN :
978-1-4244-5272-9
DOI :
10.1109/CNMT.2009.5374764