Title: Temporal pattern recognition based interactive video-on-demand streaming technique

Authors: Jun Pyo Lee; Sang Hee Kim; Young Woo Park

Addresses: Software R&D Lab., LIG Nex1, 702, Sampyeong-dong, Bundang-gu, Seongnam-city, Gyeonggi-do, Korea ' The 3rd R&D Institute-1, Agency for Defence Development, 160, BukYuseongDaero, 488, Yoseong-gu, Daejeon, Korea ' Software R&D Lab., LIG Nex1, 702, Sampyeong-dong, Bundang-gu, Seongnam-city, Gyeonggi-do, Korea

Abstract: In this paper, we propose an interactive video-on-demand (VoD) streaming technique using the temporal pattern recognition for an efficient utilisation of video proxy server. The storage management method of video proxy server is based on probabilistic parameters of a Markov model for selective saving of often-used video data. At the first step, the request orders of video data are sequentially identified and the bi-gram transition probability (BTP) of video is calculated using request orders. At the second step, the total probability or base probability is calculated using the created state transition model. Finally, the decision probability is calculated using the video data on the input queue when the storage size is insufficient. If the base probability is lower than the decision probability, we select the video data with the lowest conditional probability as a removal object in order to free up storage space of video proxy server.

Keywords: temporal pattern recognition; interactive streaming; video proxy server; probability model; video-on-demand; VoD; modelling; storage space.

DOI: 10.1504/IJAMC.2014.060502

International Journal of Advanced Media and Communication, 2014 Vol.5 No.2/3, pp.149 - 164

Received: 07 Jun 2013
Accepted: 05 Nov 2013
Published online: 17 Apr 2014 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article