Inter-object layer clustering for scalable video streaming

Hyunjoo Kim, Heon Y. Yeom, Sooyong Kang*, Youjip Won

*Corresponding author for this work

Research output: Contribution to journalArticle

Abstract

In this work, we develop an efficient storage technique to support realtime streaming of layer encoded video in a single hard disk. The size of a single hard disk drive will soon be able to hold multi-tera bytes and is going to handle relatively larger number of files. We expect that disk layout in a single disk will be rather critical issue in determining the efficiency of the storage system.We propose a novel storage technique, Inter-Object Layer Clustering for layer encoded video objects. In Inter- Object Layer Clustering, storage is partitioned into two regions: lower layer partition and upper layer partition. Lower and upper layer partition harbor the lower layer and upper layer data blocks across all video objects and cluster them together. We develop an elaborate performance model for this placement scheme. We examine the performance of the proposed technique using analytical formulation as well as a physical experiment. We found that clustering the layers across all objects brings 100% increase in the number of concurrent sessions compared to the case where file is stored in temporal order when the clients' access bandwidth is narrow. Inter-Object Layer Clustering shows 15% performance improvement compared to the clustering of layers within the objects.

Original languageEnglish
Pages (from-to)313-333
Number of pages21
JournalMultimedia Tools and Applications
Volume50
Issue number2
DOIs
Publication statusPublished - Nov 2010
Externally publishedYes

Keywords

  • Data placement
  • File system
  • Layered encoding
  • Scalable video
  • Video streaming

Fingerprint Dive into the research topics of 'Inter-object layer clustering for scalable video streaming'. Together they form a unique fingerprint.

Cite this