Loading in 2 Seconds...
Loading in 2 Seconds...
Coordinate Live Streaming and Storage Sharing for Social Media Content Distribution. Authors: Xu Cheng, Jiangchuan Liu , Senior Member, IEEE , Haiyang Wang, and Chonggang Wang , Senior Member, IEEE. Speaker : 童耀民 MA1G0222. 2013.12.23. Outline. INTRODUCTION MEASUREMENT STUDY
Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.
Coordinate Live Streaming and Storage Sharing for Social Media Content Distribution Authors:Xu Cheng, Jiangchuan Liu, Senior Member, IEEE, HaiyangWang, and Chonggang Wang, Senior Member, IEEE Speaker :童耀民 MA1G0222 2013.12.23
Outline • INTRODUCTION • MEASUREMENT STUDY • USER QUESTIONNAIRE SURVEY • COOLS SYSTEM OVERVIEW • COOLS DESIGN DETAILS • IMPROVING COOLS OVERLAY TREE • PERFORMANCE EVALUATION • CONCLUSION AND FUTURE WORK
INTRODUCTION • The recently emerged user-generated contents (UGC) services, social networking services (SNS), as well as the pervasive wireless mobile network services have formed social media which has drastically changed the content distribution landscape.
INTRODUCTION • Today such UGC applications as YouTube allow any user to be a content provider, generating enormous amount of video contents that are quickly and extensively propagated on the Internet through such SNSes as Facebook and Twitter.
INTRODUCTION • Unfortunately, the sheer and ever-increasing data volume, the broader coverage, and longer access durations of video objects also present significant challenges than other types of objects, not only to the social networking website management, but also to the network traffic engineering and to the resource provisioning of external video sites.
INTRODUCTION • To better understand the challenges and opportunities therein, they investigate the social networking users’ behavior from both system traces and a questionnaire survey.
INTRODUCTION • Through the posting function of social networking services, the server can broadcast the live video to the user’s friends, who can be either wired Internet users or mobile users. • 1)a friend can choose to watch the live video, and thus the requirement of streaming quality, such as startup latency and playback continuity, should be satisfied;
INTRODUCTION • 2) a friend can choose not towatch the live video, but she or he can download the video and expect to watch it later. • 3) a friend shows no interest in the video.
INTRODUCTION • In this paper, they present Coordinated Live Streaming and Storage Sharing (COOLS), a system for efficient peer-to-peer posting of user-generated videos. • Through a novel ID code design that embeds nodes’ locations in a tree overlay. • They also improve our overlay tree to achieve better efficiency and robustness.
MEASUREMENT STUDY • Fig. 2. Number of watched and shared videos against rank.
MEASUREMENT STUDY • Fig. 3. CDF of time span from post to watch.
USER QUESTIONNAIRE SURVEY • Fig. 4. Breakdown of user’s concern on videos.
USER QUESTIONNAIRE SURVEY • Fig. 5. Comparison of the possibility of watching the entire video.
COOLS SYSTEM OVERVIEW • Fig. 6. Example of overlay tree with ID.
COOLS DESIGN DETAILS Fig. 8. Example of node demotion.
IMPROVING COOLS OVERLAY TREE • We present a novel improved overlay tree structure: if the root node has children , then the nodes at depth have at most children, and the tree height is no greater than .
IMPROVING COOLS OVERLAY TREE Fig. 9. Example of improved overlay tree.
IMPROVING COOLS OVERLAY TREE • To understand this, we assume a complete tree in which the root has 4children, and thus there are at most20 nodes and the tree height is 3 . • Given that a complete tree in which the root nodehas children, there are nodes at depth 1, nodes at depth 2, and so on.
IMPROVING COOLS OVERLAY TREE source Root 4=2k; k=2 Root Root Root Root d1 d1 d1 d1 d1 d1 d1 d1 d2 d2 d2 d2 d2 d2 d2 d2
PERFORMANCE EVALUATION Fig. 10. CDF of startup delay.
PERFORMANCE EVALUATION Fig. 11. CDF of data loss rate
PERFORMANCE EVALUATION Fig. 12. Comparison of overhead size
CONCLUSION AND FUTURE WORK • Thanks.